Cloning, Assembly, and Modification of the Primary Human Cytomegalovirus Isolate Toledo by Yeast-Based Transformation-Associated Recombination

Compared to other primary isolates, BAC-cloned HCMV Toledo displayed an inversion in the so-called ULb’ region that is frequently mutated upon multiple passages of HCMV. Specifically, BAC-cloned Toledo displayed an inversion resulting from a recombination event within sequences located between UL128 and UL129 at one end and UL133 and UL148a at the other end (Fig. 1A) (11). This recombination event led to a truncation of UL128 (19), a chemokine-like protein that is a subunit of the pentameric complex gH/gL/UL128/UL130/UL131A (20, 21). While the pentameric complex promotes infection of endothelial, epithelial, and myeloid cells, mutations in the nonessential subunits UL128, UL130, and UL131A are selected for upon passaging in fibroblasts, presumably due to the fact that pentamer-deficient viruses are less cell associated (12). Since pentameric complex mutants of nonhuman primate CMVs are attenuated in vivo (22), this fibroblast adaptation was at odds with the clinical observations in volunteers inoculated with the Toledo isolate. Therefore, we determined whether early passages of Toledo already contained this deletion by isolating DNA from supernatants of fibroblasts infected with Toledo passage 7 (P7), the earliest passage still available, and performing diagnostic PCR of genes adjacent to UL128 of UL133 and the native versus inverted configuration. In the HCMV isolate TR, used as a control, UL128 is adjacent to UL127, whereas UL133 is adjacent to UL148 as indicated by the appropriate PCR fragments PCR-1 and PCR-3 (Fig. 1B). Interestingly, the same fragments were also amplified from P7 Toledo, indicating a noninverted genome orientation similar to that of TR. In addition, P7 Toledo also contained genomes in the inverted configuration in which UL128 is adjacent to UL148 (PCR-4) and UL133 is adjacent to UL127 (PCR-2) (Fig. 1B). Thus, the early passages of Toledo used in clinical trials consisted of a mixture of wild-type and fibroblast-adapted genome configurations.

To further examine the genome content of low-passage-number Toledo samples in an unbiased fashion, we employed Illumina HiSeq paired-end sequencing on viral DNA isolated from the supernatant of fibroblasts infected with P7 Toledo. The depth of sequence coverage was sufficient to obtain the full-length sequences of both the majority sequence representing the fibroblast-adapted Toledo strain and the minority, presumably parental, wild-type sequence. We will refer to the fibroblast-adapted Toledo strain and the presumably parental wild-type strain as Toledo-F and Toledo-P, respectively. Toledo-F was mostly identical to the Toledo sequence deposited into GenBank (GenBank accession number GU937742.2 ). However, a few single nucleotide polymorphisms (SNPs) were identified, most of them synonymous mutations (see Fig. S1 in the supplemental material). These SNPs are either due to sequencing errors in the GenBank sequence or mutations that occurred during further passaging and cloning of the Toledo isolate. In order to model the sequence of the Toledo-P genome and the Toledo-F genome, we edited the reference genome sequence (GenBank accession number GU937742.2 ) based on our sequencing results. Individual SNPs were allocated to either one or both Toledo genomes based on the following criteria. If the variation was homogeneous, it was applied to both genomes. If the variation was heterogenous, the major allele was applied to the fibroblast-adapted genome, while the minor allele was applied to the parental genome. Usually, for heterogeneous variations, one of the alleles would match the allele in the reference genome, such that “applying” it would be to take no action to the derivative genome at that locus. Alignment of the full-length sequences of Toledo-F and Toledo-P revealed that, in addition to the inversion of the UL128-UL133 fragment which is flanked by two small deletions, Toledo-F had a 279-bp in-frame deletion in the RL13 gene, as previously reported (23), a 59-bp insertion in the intergenic region between UL59 and UL60, as well as several SNPs in the UL region (Fig. 2A and B). By comparing the sequence coverage (number of reads) of Toledo-F versus Toledo-P at the SNP loci, we were able to estimate the ratio of the two sequences in the viral DNA preparation. The P7 Toledo-F genomes represented the vast majority of sequences, whereas only 5% of the reads belonged to Toledo-P (Fig. 2C). Since only these two genome sequences were detected in the P7 sample, it is highly likely that P7 represents an intermediate mixture in which most of the culture has been taken over by a faster-growing, fibroblast-adapted variant, with some of the original isolate remaining. Similar observations have been reported upon serial passaging of the primary isolate Merlin where mutations in RL13 occurred within three passages, and mutations in the UL128-to-UL131A (UL128-131A) region upon subsequent passages (12). Since clinical studies were performed with passage 4 Toledo, these data also suggest that the percentage of non-fibroblast-adapted, wild-type, sequence was likely higher, although no passage 4 samples remain, making it impossible to determine the exact composition.

The observed presence of different variants within a given primary isolate illustrates a typical challenge for the cloning of HCMV genomes. Continued passage of Toledo likely resulted in the presence of only the Toledo-F population so that two independently derived BAC clones correspond to Toledo-F (11, 24). Moreover, four different chimeras of HCMV Toledo cloned as overlapping cosmids with the vaccine strain HCMV Towne were exclusively based on Toledo-F which might explain their limited infectivity and immunogenicity in clinical studies (25–27). To enable the cloning of primary isolates with limited propagation in tissue culture, we therefore used the P7 Toledo sample as a test case for a new TAR cloning-based strategy applicable to HCMV isolates in general (Fig. 3). By aligning a number of known HCMV isolates, we identified highly conserved sequences that were used to design 40-bp-long, synthetic “TAR hooks” (Table S1) for the cloning of the entire genome as 16 overlapping fragments (Table S2). Pairs of adjacent TAR hooks flanking 10- to 20-kb regions were inserted into a yeast centromeric plasmid with BAC sequence (YCpBAC) and transformed into yeast together with sheared P7 HCMV Toledo DNA (Fig. S2A). Each of the 16 fragments which contain 80 bp of homology between adjacent fragments as well as flanking I-SceI sites was characterized by PCR and restriction analysis as shown for fragment TAR04 (Fig. S2B to D). The efficiency for cloning the HCMV fragments ranged from 20% (TAR15) to 80% (TAR03). Since multiple attempts to obtain full-length genomes from the 16 overlapping DNA fragments failed, we assembled the full-length HCMV Toledo in two steps. In the first step, we generated two half genomes by cotransforming overlapping DNA fragments 1 to 8 and fragments 9 to 16 together with the YCpBAC vector into S. cerevisiae (Fig. 4A). The resulting half genomes were also transferred into E. coli and characterized by PCR analysis (Fig. 4A). Assembly of the half genomes was very efficient. Positive results were observed in 95% of the clones that were screened. To generate a full-length genome, the half genomes were released from the YCpBAC by I-SceI or PacI restriction digestion and transformed together with linearized YCpBAC containing TAR hooks for the beginning and end of the HCMV genome (Table S1). The resulting full-length genomes in S. cerevisiae were analyzed by junction PCR analysis (Fig. 4B), and positive clones were transformed into E. coli. The success rate for assembling full-length Toledo-F genomes in yeast was approximately 15%, and the transfer of the full-length genomes to E. coli was close to 100% efficient. Next-generation sequence analysis showed that the TAR-assembled HCMV genome was identical to the P7 Toledo-F sequence present in the low-passage-number sample (Fig. 4C).

To assemble the Toledo-P genome also, we took advantage of the ability to simultaneously change the sequence of each of the 16 fragments either by mutagenesis or by gene synthesis. The major differences between Toledo-P and Toledo-F are located on three fragments: fragment TAR01 contains the RL13 sequence, fragment TAR07 contains open reading frames (ORFs) 59 and 60, and fragment TAR13 contains the UL128-UL133 (UL128-133) region. There were also four SNP differences between Toledo-P and Toledo-F (Fig. 2). However, since these either occurred in noncoding regions or were synonymous, we did not modify them in our assembly of Toledo-P.

TAR01 and TAR07 corresponding to the Toledo-P sequence were generated by in vitro Cas9 digestion followed by assembly with a PCR fragment containing the Toledo-P sequence (Fig. 5A). TAR13 was modified to contain the appropriate Toledo-P sequence by first amplifying the inverted region with primers that not only contained the additional sequences for Toledo-P but also contained sequences to restore the proper orientation of the UL128-133 region. A YCpBAC with the remaining Toledo sequences was also amplified, and the two fragments were assembled utilizing Gibson Assembly (SGI-DNA) and transformation into E. coli.

The resulting TAR plasmids were sequenced to confirm that the respective Toledo-P fragments had been generated (data not shown). The full-length Toledo-P genome was then generated from two half genomes in which Toledo-P fragments TAR01, TAR07, and TAR13 replaced the corresponding Toledo-F fragments. The assembly of the full-length Toledo-P genome in yeast and transfer to E. coli were not as robust as those of Toledo-F. Only 3% of the yeast Toledo-P clones were correct (compared to 15% for Toledo-F), and the efficiency of Toledo-P full-length genome transfer to E. coli was 10% (compared to almost 100% for Toledo-P). The reasons for this relative inefficiency for Toledo-P are not yet understood. The resulting full-length YCpBAC containing Toledo-P genome was analyzed by next-generation sequencing (NGS) (Fig. 5B and C). Except for the four SNPs that were not repaired (and hence occur in 100% of the assembled virus), the sequence of the assembled full-length Toledo-P YCpBAC corresponded to the minority, Toledo-P sequence in the P7 Toledo sample.

The TAR-cloned HCMV Toledo genomes were released from the YCpBAC by restriction enzyme digestion targeting the flanking restriction sites with PacI endonuclease. To recover virus, the released full-length genomes were transfected into MRC-5 fibroblasts. After 4 passages in MRC-5 cells, purified viral stocks were generated for in vitro and in vivo characterization. At the same time, viral DNA was extracted by the Hirt method and sequenced by NGS. Comparison of the Toledo-P virus with the YCpBAC sequence revealed frameshift mutations in RL13 amino acid residue 193 and in UL128 amino acid residue 154 (Fig. 5B and C). The frameshift mutation in RL13 and UL128 were found in approximately 71% and 85% of the genomes, respectively, suggesting that mutations in these gene regions are rapidly selected for even when starting with a molecular clone. However, while the RL13 mutation truncates the protein by 50% and likely inactivates the protein, the frameshift in UL128 eliminates only the final 16 amino acids. Since UL128 is a subunit of the pentameric complex that facilitates entry into nonfibroblast cells, we determined whether the UL128 truncation would affect the anticipated ability of Toledo-P to infect endothelial cells. In a multistep growth curve (multiplicity of infection [MOI] of 0.01) in MRC-5 cells, both Toledo-P and Toledo-F generated comparable titers of virus in the supernatant and with comparable kinetics (Fig. 6A). In contrast, only Toledo-P was able to grow in human umbilical cord endothelial cells (HUVECs), whereas Toledo-F did not grow (Fig. 6A). These data suggest that Toledo-P retained a functional pentameric complex despite the truncation in UL128.

We previously reported a humanized mouse model in which primary isolates of HCMV are able to establish latency and can be reactivated upon mobilization of myeloid cells from the bone marrow (28–30). In contrast, laboratory-adapted strains such as AD 169 are unable to do so (P. Caposio, unpublished observations). To compare Toledo-P and Toledo-F in this model, we engrafted NOD-scidIL2Rγc null (NSG) mice with CD34⁺ human progenitor cells. At 12 to 14 weeks after engraftment, mice were infected via intraperitoneal injection of human fibroblasts infected with HCMV TR, HCMV Toledo-P, or HCMV Toledo-F. At 8 weeks postinfection, the mice were split into two groups, and one group was treated with granulocyte colony-stimulating factor (G-CSF) and AMD3100 (see Materials and Methods) was administered intraperitoneally to mobilize human monocytes/macrophages. At 1 week after mobilization, the mice were sacrificed. Total DNA was extracted from the spleens and livers, and HCMV genomes were analyzed using quantitative PCR. When comparing G-CSF-treated to untreated groups, we were able to detect quantitative differences in viral DNA loads in the spleen and liver for both HCMV TR and Toledo-P (Fig. 6B). In contrast, no increase in viral genome copies were observed for Toledo-F. Moreover, compared to HCMV-TR and Toledo-P, genome copy numbers tended to be lower even in nonmobilized mice infected with Toledo-F. Thus, Toledo-P and Toledo-F strongly differ in their ability to establish latency and reactivate in vivo. Only Toledo-P behaves similarly to other primary isolates (TR, TB40E) in this model, whereas Toledo-F resembles laboratory-adapted strains like AD 169.

Human primary embryonic lung fibroblasts (MRC-5; purchased from ATCC, Manassas, VA) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) and supplemented with 10% fetal bovine serum (FBS), 10 mM HEPES, 1 mM sodium pyruvate, 2 mM l-alanyl-glutamine, 0.1 mM nonessential amino acids, 100 U/ml penicillin, and 100 µg/ml streptomycin. Human umbilical vein endothelial cells (HUVECs) were purchased from Lonza (cc-2519) and cultured in endothelial growth medium (EGM-2; Clonetics) containing 2% FBS, human recombinant vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), human epidermal growth factor (hEGF), insulin-like growth factor 1 (IGF-1), hydrocortisone, ascorbic acid, heparin, gentamicin, and amphotericin B (1 μg/ml each) (high-serum medium). HCMV isolate TR was derived originally from a patient with AIDS-related HCMV retinitis (39). HCMV Toledo-P and Toledo-F viral DNA was extracted from E. coli using a Qiagen Midi-kit, digested with PacI, and electroporated into MRC-5 cells. Infectious virus was produced, and the titers of the virus were determined by plaque assay on MRC-5 cells as previously described (40).

Plasmids pCC1BAC-ura3 and pCC1BAC-his3 have been previously described (41).

Supernatant from fibroblasts infected with Toledo isolate passage 7 (P7) was used to extract viral DNA using the QIAamp MinElute virus spin kit (Qiagen). HCMV TR DNA was extracted with the same kit as a positive control. Two microliters of viral DNA was amplified with Taq 5× Master Mix (New England BioLabs [NEB]) using the following primers (indicated in the parentheses) for the regions and diagnostic PCR indicated: UL127 (5′-ATGTGCCAGCTTGATGTCGC-3′) and UL130 (5′-CGCCAAGATTTTTGGAGCGCAC-3′) (PCR-1); UL127 (5′-ATGTGCCAGCTTGATGTCGC-3′) and UL133 (5′-GGTTGTGAACTCACCGTCGG-3′) (PCR-2); UL133 (5′-GGTTGTGAACTCACCGTCGG-3′) and UL148 (5′-CGAGGCAGAACATCTCAACC-3′) (PCR-3); UL128 (5′-GAGGGCCTTACAGCCTATGG-3′) and UL148 (5′-CGAGGCAGAACATCTCAACC-3′) (PCR-4). The thermocycler parameters were 95°C for 2 min, followed by 30 cycles, with 1 cycle consisting of 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min.

Viral DNA was isolated from the supernatant of fibroblasts infected with P7 Toledo as described above, and 210 μl of isolated DNA was incubated with 30 μl of shrimp DNase to remove host DNA (2 U/μl) (Affymetrix) for 1 h at room temperature. Then, the shrimp DNase was inactivated by incubating at 70°C for 20 min. A 10-μl sample was removed for quantitative PCR (qPCR) analysis. Viral DNA was then purified from the larger remaining sample with the QIAamp MinElute virus spin kit (Qiagen) and eluted in 50 μl of buffer. qPCR was used to determine the removal of host DNA. Viral DNA was then used for sequencing using Illumina HiSeq paired-end sequencing. The sequence reads were sorted by barcode, trimmed, and mapped to HCMV Toledo (GenBank accession number GU937742.2 ) using CLC Bio’s (Qiagen) clc_ref_assemble_long program, and variations were then extracted and reported using CLC Bio’s find_variations program. Reported variations were manually inspected using CLC Bio’s assembly_viewer program, and variations lacking sufficient evidence (lack of spanning reads or insufficient high-quality coverage) were discarded. In some cases, large structural variations were analyzed and reported manually, for example, a large inversion with indels at the termini that occurred at low frequency had to be determined by bespoke bioinformatic methods to partition sequencing reads based on the presence or absence of junction sequences, and then reassembling the partitions of reads separately and mapping contigs back to the reference sequence.

The P7 Toledo-F, P7 Toledo-P, Assembled Toledo-F, Reconstituted Toledo-F, Assembled Toledo-P, and Reconstituted Toledo-P were aligned in Geneious (Biomatters Ltd., Auckland, New Zealand) using the PhyML package to the reference strain (HCMV Toledo, GenBank accession number GU937742.2 ) available from NCBI and to each other for identification of insertions, deletions, and single nucleotide polymorphisms (SNPs).

The S. cerevisiae strain VL6-48N (14) (MATα his3-Δ200 trp1-Δ1 ura3-Δ1 lys2 ade2-101 met14 cir°) was used for all yeast transformations and grown in YPD (yeast extract, peptone, dextrose) medium supplemented with adenine. S. cerevisiae spheroplasts were prepared using previously described methods with the following modifications (16). Overnight cultures were grown to an optical density at 600 nm (OD₆₀₀) of 1.8 to 2.5 and kept overnight at 4°C in 1 M sorbitol. After the cells were harvested, they were resuspended in 10 ml of SPE solution (1 M sorbitol, 0.01 M sodium phosphate, 0.01 M Na₂EDTA) with 20 μl of β-mercaptoethanol and 20 μl of zymolyase solution. The cells were finally resuspended in 4 ml of STC solution (1 M sorbitol, 0.01 M Tris-HCl, 0.01 M CaCl₂). DNA for transformation-associated recombination (TAR) cloning or assembly was added to 200 μl of spheroplasts, followed by the addition of 900 μl of 20% PEG solution (20% PEG 8000, 10 mM CaCl₂, 10 mM Tris-HCl). The suspension was incubated at room temperature for 20 min, the PEG solution was removed, and the cells were then incubated at 30°C for 30 min in SOS medium (1 M sorbitol, 6.5 mM CaCl₂, 0.25% yeast extract, 0.5% peptone). Transformed yeast spheroplasts were plated with selective top agar with sorbitol.

Vectors were first PCR amplified using pCC1BAC-his3 as the template with Phusion (NEB) or Q5 (NEB) DNA polymerase. This plasmid contains bacterial artificial chromosome (BAC) and yeast centromeric plasmid (YCp) sequences for replication in E. coli and yeast. The primers add an I-SceI or PacI restriction site, flanked by 40 bp of HCMV homology, to each end of the vector backbone. PCR products were digested with DpnI (NEB) prior to transformation.

Viral DNA, which was isolated from Toledo P7 as described above, was sheared by pipetting. Six hundred nanograms of viral genomic DNA was cotransformed with 20 ng of the appropriate vector into yeast spheroplasts. Transformants were patched on synthetic dropout medium without histidine and supplemented with adenine (SD-HIS) plates and after sufficient growth were selected and transferred into 20 μl of 25 mM NaOH and incubated at 95°C for 30 min to lyse yeast cells. PCR on DNA from lysed yeast using the appropriate detection primers (see Table S1 in the supplemental material) was used to confirm the correct junction of the vector and HCMV fragment on each side.

Positive TAR clone candidates from yeast patches or 3 ml of liquid cultures were resuspended in 500 μl of water containing 5 μl of zymolyase 20T (MP Bio) (10 mg/ml) and 0.5 μl of β-mercaptoethanol (14.2 M) and incubated at 37°C for 1 h. Fifty microliters of 2% SDS was added, and the solution was incubated for 15 min at 70°C, followed by the addition of 50 μl of 5 M potassium acetate and incubation on ice for 5 min. After clarification, the DNA in the supernatant was precipitated with isopropanol and resuspended in 50 μl of Tris-EDTA (TE) buffer. Isolated DNA was then electroporated into E. coli EPI300 (Epicentre) or DH10B (Thermo Fisher) competent cells. DNA was purified from positive transformants for restriction enzyme analysis and sequencing.

Cultures of all 16 TAR clones in E. coli were grown up, and DNA was isolated with PureLink HiPure Plasmid Midiprep kit (Thermo Fisher). Prior to assembly, the clones were digested with I-SceI and heat killed to release the HCMV fragments. Assembly of HCMV genomes was performed by TAR assembly in yeast in two steps. First, half genomes (TAR01 to TAR08 and TAR09 to TAR16) were assembled as follows. Vector DNA was amplified by PCR using pCCIBAC-ura3 as the template and either Con01R (R stands for reverse)and Con08F (F stands for forward) or CMV_1R and CMV_8F (Table S1) as primers for TAR01 to TAR08 or Con09R and Con16F or CMV_9R and CMV_16F (Table S1) for TAR09 to TAR16. For assembly of each half genome, 10 ng of vector DNA and 100 ng each of all other DNA fragments was added to spheroplasts. Transformants were patched on plates containing synthetic dropout medium without uracil and supplemented with adenine, and DNA was screened by PCR for the appropriate junctions with the respective detection primers (Table S1) using NaOH lysis. Positive half genomes were transformed into E. coli EPI300 competent cells and screened by PCR to confirm the appropriate junctions. For assembly of the full genome, the half genomes were processed with I-SceI or PacI endonuclease as described above to release the fragments. Vector DNA was amplified by PCR using pCCIBAC-his3 as the template and Con01R and Con16F or CMV_1R and CMV_16F as primers (Table S1). Ten nanograms of vector DNA and 100 ng each of the half genomes was added for assembly to spheroplasts. Transformants were patched onto SD-HIS plates, and DNA was screened by PCR for all of the appropriate junctions with the respective detection primers (Table S1) using NaOH lysis methods previously mentioned. As before, positive clones were transformed into E. coli EPI300 competent cells as described above and screened by PCR to confirm the appropriate junctions.

A single colony of E. coli Alpha-Select Gold Efficiency (Bioline) harboring the pJExpress404-FLAG-Cas9-His construct was cultured overnight in 50 ml LB supplemented with 100 μg/ml carbenicillin (Teknova). The starter culture was diluted into 1 liter of antibiotic-supplemented LB broth and shaken at 220 rpm at 37°C to an OD₆₀₀ of 0.6. The culture was then cooled to 20°C, and isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM to induce protein expression. Cas9 protein was induced overnight in a refrigerated incubator at 20°C under shaking (220 rpm). Cell pellets were washed once with phosphate-buffered saline (PBS), pH 7.4, and then resuspended in 60-ml volumes of xTractor buffer supplemented with EDTA-free protease inhibitor (Clontech), 2 mg lysozyme (Sigma-Aldrich), and 625 U Benzonase (Millipore). Cell suspensions were then gently rotated for 2 h at 4°C prior to mechanical lysis in a Microfluidizer M-110L (Microfluidics). Cell debris was pelleted at 8,000 × g for 1 h, and recombinant protein was purified from the supernatant using a HisTalon column (Clontech) according to the manufacturer’s instructions. Purified protein was stored at −80°C in Cas9 buffer (150 mM KCl, 20 mM HEPES [pH 7.4], 1 mM dithiothreitol [DTT], 10% glycerol) at a concentration of 4.5 mg/ml. The Cas9-targeting chimeric synthetic guide RNAs (sgRNAs) were designed, in vitro synthesized, and purified as described previously (42).

Previously assembled HCMV Toledo-F fragments TAR01 and TAR07 contained within the E. coli-yeast shuttle pCC1BAC_HIS3 were purified using the QIAprep Spin Miniprep kit (Qiagen). Plasmid-borne genomic fragments were then independently digested with Cas9 in conjunction with two sgRNAs (fragment 1, 5′-CAAUGGACUGGCGAUUUAGUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCUUUUUUU-3′ and 5′-GAUGCGAUCGCAGUUACGGUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCUUUUUUU-3′ and fragment 7 5′-UGUGGAAUUCCGGACAUAGUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCUUUUUUU-3′ and 5′-UUACGUAUACCGGAUGCUGUUUUAGAGCUAGAAAUAGCAAGUUAAAAUAAGGCUAGUCCGUUAUCAACUUGAAAAAGUGGCACCGAGUCGGUGCUUUUUUU-3′ [Cas9 targeting sequences are underlined]), liberating DNA fragments of roughly 350 bp and 70 bp, respectively. One-hour digestions were performed at 37°C in 50-μl volume and contained 500 ng of plasmid DNA, 1 μg each gRNA, 10 mM spermidine (Sigma-Aldrich), 10 mM MgCl₂ (Ambion), 20 mM HEPES, 150 mM KCl, 0.5 mM DTT, 0.1 mM EDTA, and 3.6 μg of Cas9 protein. Following digestion, Cas9 was heat inactivated at 80°C for 10 min, and digested DNA was extracted with UltraPure phenol-chloroform-isoamyl alcohol (25:24:1, vol/vol; pH 8), precipitated with ethanol, and resuspended in sterile water. DNA fragments used to generate Toledo-P insert fragments 1 and 7 were obtained from de novo-synthesized gblocks (IDT). The Toledo-P fragment 1 insert encoded a 279-bp insertion not found in the Toledo-F sequence. Conversely, the Toledo-P fragment 7 insert removed a 52-bp insert found only in Toledo-F. Both DNA inserts contained 20- to 100-bp overlap sequences homologous to regions upstream and downstream of the Cas9 digestion sites to facilitate in vitro assembly. Synthesized insert fragments were mixed with corresponding digested products at a 3:1 molar ratio (insert-vector) and covalently joined via standard Gibson Assembly reaction (SGI-DNA). Following assembly, constructs were transformed into E. coli TransforMax EPI300 electrocompetent cells (Epicentre) and plated on selective media. Transformants were verified to contain new inserts using PCR and Sanger sequencing prior to final confirmation of the full Toledo-P genome sequence by next-generation sequencing (NGS).

NGS sequencing data indicated that the Toledo-P genome contained a large 14.3-kb genomic inversion flanked by two small insertions (14 and 23 bp). This large inversion was entirely contained within the previously generated Toledo-F genomic fragment 13. Utilizing primers 5′-GCAAAGTGAACGACAAGGCGCAGTACCTGCTG-3′ and 5′-CCTAGTAACACTCGTCCGACACTTCCACCATCTCCAGC-3′ (IDT), the ends of genomic fragment 13 backbone were PCR amplified together with the E. coli-yeast shuttle pCC1BAC_HIS3 vector. In a second PCR, the 14.3-kb inverted sequence was amplified with Ultramer oligonucleotides 5′-CTCTCCAGGTACTGATCCAGGCCCACGATCCGGGTTATCTTGTCGTATTCCAGGTTGATCCATCGATAGGGAACGCTGCCAGCGGCGCCCAGCAGGTACTGCGCCTTGTCGTTCACTTTGCCGCAGCGTATTCGCCCGTCAGCTTCGAGGTATAACCTACAACACGGAGGGGAAGGGGGGTACAAAACGTGAAATTAGAC-3′ and 5′-GAGACGACGCCGCTGGTAGAGGATGCCGAACCGCCGGCCGAGCTGGAGATGGTGGAAGTGTCGGACGAGTGTTACTAGGAGATCGCCGCGGCCGATGGGCGCCGGCGGACGTGACTCGGCAGCCGCTGTAGGGATAAATAGTGCGATGGCGTTTGTGG-3′, generating an inverted sequence with the indicated terminal inserts flanked by 121-bp overlap sequences homologous to regions at the ends of the first amplicon. Both PCRs were performed with the KOD Xtreme Hot Start DNA polymerase (Millipore) using the Toledo-F genomic fragment 13 as the template and according to the manufacturer’s specifications. The amplicons were digested with DpnI (NEB) and purified using the QIAquick PCR purification kit (Qiagen). Following purification, amplicons were mixed at an equal molar ratio and covalently joined via standard Gibson Assembly reaction (SGI-DNA). Following assembly, constructs were transformed into E. coli TransforMax EPI300 electrocompetent cells (Epicentre) and plated on selective media. Transformants were verified to contain new inserts using PCR and Sanger sequencing prior to final confirmation of the full Toledo-P genome sequence by NGS.

All animal studies were carried out in strict accordance with the recommendations of the American Association for Accreditation of Laboratory Animal Care (AAALAC). The protocol was approved by the Institutional Animal Care and Use Committee (number IS00003498) at the Oregon Health & Science University. NOD-scid IL2Rγ_c^null (NSG) mice were maintained at a pathogen-free facility at Oregon Health & Science University in accordance with procedures approved by the Institutional Animal Care and Use Committee. Both sexes of animals were used. Humanized mice were generated as previously described (30). The animals (12 to 14 weeks after engraftment) were treated with 25 ng/mouse lipopolysaccharide (LPS), and after 6 h, they were infected via intraperitoneal injection of human dermal fibroblasts previously infected with HCMV TR, Toledo-P, or Toledo-F at approximately 5 × 10⁵ PFU per mouse. A control group of engrafted mice were mock infected using uninfected fibroblasts. At 4 weeks postinfection, the infected mice were split into two groups, and one group was treated with 100 μl of granulocyte colony-stimulating factor (G-CSF) (300 mg/ml; Amgen) using a subcutaneous micro-osmotic pump (1007D; Alzet) and 125 μg AMD3100 [1,1′-[1,4-phenylenebis(methylene)]bis-1,4,8,11-tetra-azacyclotetradecane octahydrochloride or plerixafor] administered intraperitoneally to mobilize hematopoietic progenitor cells (HPCs). The remaining mice serve as a direct comparison for the effects of virus reactivation and dissemination that follow HPC mobilization. One week after HPC mobilization, the mice were sacrificed, their organs were harvested, and samples for PCR were frozen in RNAlater solution and stored at −80°C for subsequent analysis.

Total DNA was extracted from approximately 1-mm² sections of mouse spleen or liver using the DNAzol kit (Life Technologies). HCMV genomes were analyzed using quantitative PCR (TaqMan) performed on 1 µg of total DNA and using TaqMan FastAdvance PCR master mix (Applied Biosystems) according to the manufacturer’s instructions. Primers and a probe recognizing HCMV UL141 were used to quantify HCMV genomes (probe, CGAGGGAGAGCAAGTT; forward primer, 5′-GATGTGGGCCGAGAATTATGA; reverse primer, 5′-ATGGGCCAGGAGTGTGTCA). The probe contains a 5′ FAM (6-carboxyfluorescein) reporter molecule and a 3′ quencher molecule (Applied Biosystems). The reaction was initiated using TaqMan Fast Advanced master mix (Applied Biosystems) activated at 95°C for 10 min followed by 40 cycles (1 cycle consists of 15 s at 95°C and 1 min at 60°C) using a StepOnePlus TaqMan thermocycler. Results were analyzed using ABI StepOne software. Data were analyzed using the statistical program GraphPad Prism 5. Statistical significance was determined using two-way analysis of variance, followed by Bonferroni’s posttest correction. A P value of <0.05 or lower was considered significant.

The genome sequences of P7 Toledo-F, P7 Toledo-P as well as assembled Toledo-F and Toledo-P have been submitted to GenBank under accession numbers MF783090 to MF783093 .