Comparison of electron microscopic techniques for enumeration of endogenous retrovirus in mouse and Chinese hamster cell lines used for production of biologics.

Murine myeloma and Chinese hamster ovary cells are used widely in the manufacture of recombinant proteins for biopharmaceuticals. However, rodent cell lines express endogenous retrovirus, which necessitates appropriate design of purification processes to remove virus in excess of the calculated maximum retroviral load. Currently, electron microscopy is the method of choice for determination of retroviral titre in bulk harvest. In this study we compared three electron microscopy techniques to determine retroviral titre in bulk harvest. These were direct negative stain, negative stain after sucrose-density purification and thin section electron microscopy of pelleted supernatant. The study demonstrated that the level of C-type retrovirus associated with cells was predictive of the viral load in cell culture supernatants. The most accurate method for quantifying viral load was direct counting, followed by thin section of pelleted supernatant and negative stain after sucrose concentration. The most practical method was thin section of resuspended pelleted supernatant, which gave improved detection limits.

Xer-mediated site-specific recombination at cer generates Holliday junctions in vivo.

Normal segregation of the Escherichia coli chromosome and stable inheritance of multicopy plasmids such as ColE1 requires the Xer site-specific recombination system. Two putative lambda integrase family recombinases, XerC and XerD, participate in the recombination reactions. We have constructed an E. coli strain in which the expression of xerC can be tightly regulated, thereby allowing the analysis of controlled recombination reactions in vivo. Xer-mediated recombination in this strain generates Holliday junction-containing DNA molecules in which a specific pair of strands has been exchanged in addition to complete recombinant products. This suggests that Xer site-specific recombination utilizes a strand exchange mechanism similar or identical to that of other members of the lambda integrase family of recombination systems. The controlled in vivo recombination reaction at cer requires recombinase and two accessory proteins, ArgR and PepA. Generation of Holliday junctions and recombinant products is equally efficient in RuvC- and RuvC+ cells, and in cells containing a multicopy RuvC+ plasmid. Controlled XerC expression is also used to analyse the efficiency of recombination between variant cer sites containing sequence alterations and heterologies within their central regions.

BPV-4 induces amplification and activation of 'silent' BPV-1 in a sub-line of C127 cells.

BPV-4-induced malignant transformation of C127 mouse fibroblasts in vitro is the result of a 'hit and run' mechanism. Viral DNA is lost on continued subculture of transformed cell lines without loss of any malignant characteristics. The DNA from these cells harbours specific amplified sequences. Two such amplified fragments of approximately 10 kb and 12 kb were molecularly cloned and designated HL-10 and HL-12 respectively. HL-10 transformed C127 cells efficiently and therefore encodes a transforming function, whereas the 12 kb clone did not. Heteroduplexes showed that HL-12 was homologous to HL-10 except for two additional tandem copies of an approximately 1.7 kb sequence. Sequence analysis of HL-10 revealed that the clone contained a 5.2 kb region from BPV-1 including the transforming ORFs. Transformation studies have shown differences between HL-10 and BPV-1, indicating that the host flanking sequences may contribute to the transforming potential of the BPV-1 ORFs. The BPV-1 DNA was associated with sequences homologous to murine autonomously replicating sequences (ARS) implicated in the establishment of multiple tandem DNA repeats. As the parental cells contain the set of sequences amplified in transformed cells in single copy and show none of the characteristics of transformed cells, we conclude that BPV-4 has activated these sequences by amplification and rearrangement. These phenomena may be mediated through an interaction between BPV-4 proteins and the BPV-1 origin of DNA replication or via the ARS region.

Intrahelical pseudoknots and interhelical associations mediated by mispaired human minisatellite DNA sequences in vitro.

The human minisatellite arrays, 33.6 and 33.15, consist of tandem reiterations of a 37-nucleotide (nt) and a 16-nt repeat unit sequence, respectively, both of which contain a majority of purine bases on one strand. Knot-like tertiary structures, which mapped to the cloned arrays, were observed by electron microscopy (EM) in homoduplex molecules produced by denaturation and reannealing in vitro. They result from a primary hybridization between misaligned repeat units of the array, forming a slipped-strand structure with staggered single-stranded DNA loops, followed by a secondary hybridization between repeat units in the two loops. Depending on the relative alignment of the loops when they hybridize, a particular form of intrahelical pseudoknot is produced. Theta-shaped, figure-of-eight, and bow-shaped structures were the most common conformational isomers observed in homoduplexes flattened into two dimensions during EM preparation. At the site of a bow-shaped structure, a conformation-dependent bend of approximately 60 degrees between the flanking DNA segments is induced; the other conformations generally do not deflect the line of the main DNA axis. Paired loops, similar to the bow-shaped structure, were apically situated in some supercoiled plasmids containing the 33.6 array. Both plasmids formed intermolecular associations, consisting of two (or more) homoduplex molecules held together at or immediately adjacent to a nexus which mapped to the minisatellite sequences. These associations might arise either by interhelical hybridization between arrays or by knot-like structures interfering with branch migration of chi-form Holliday junctions.

Multiple Harvey-ras genes in the bovine genome.

Three different Harvey-ras (Ha-ras) genes have been identified in the bovine genome by screening a lambda phage-bovine DNA library with the ras gene of the Harvey murine sarcoma virus. The genes have been characterized by hybridization and heteroduplex analysis with both the viral and human Ha-ras genes. Based on heteroduplex mapping, the putative direction of transcription and the approximate position of the coding regions of the bovine genes were established. Two genes, bovine Ha-ras 1 and 2, both show an arrangement of exons and introns typical of c-Ha-ras genes. However, they are distinct entities as they have different nucleotide sequences and physical maps, and heteroduplexes between them contain a region of non-homology upstream of exon 1. The third gene, c-Ha-ras 3, does not contain introns and is a pseudogene, analogous to human c-Ha-ras 2. The nucleotide sequence of both Ha-ras 1 and Ha-ras 2 has been determined and shows that Ha-ras 1 encodes a bona fide ras protein, whereas Ha-ras 2 has diverged considerably from a recognizably functional sequence.

DNA tertiary structures formed in vitro by misaligned hybridization of multiple tandem repeat sequences.

DNA tertiary structures are shown to be formed by denaturation and reannealing in vitro of molecularly-cloned DNA containing multiple tandem repeat sequences. Electron microscopy of homoduplex DNA molecules containing the human c-Harvey-ras gene revealed knot-like structures which mapped to the position of the 812 bp variable tandem repeat (VTR) sequence. We propose that the structures result from slipped-strand mispairing within the VTR and hybridisation of homologous repetitive sequences in the single-stranded loops so produced. Similar structures were also found in freshly-linearized supercoiled plasmids. More complex knot-like structures were found in homoduplexes of a 4 kb tandem array from the hypervariable region 3' to the human alpha-globin locus. Formation of such DNA tertiary structures in vitro also provides a practical method for identifying and mapping direct tandem repeat arrays that are at least 800 bp long.

R-loop mapping of RNA transcripts from bovine papillomavirus type 4.

RNA isolated from bovine oesophageal warts was hybridised to BPV-4 genomic DNA cloned in a plasmid vector. In R-loop preparations, six major classes of transcripts were mapped. Class I, due to an unspliced transcript encompassing the E4/E5 ORF region, is most common. In Class II the E4/E5 region is spliced at its 5' end to the E6 ORF region, and the RNAs appear to have different transcription start points in the E6 ORF. Some Class I R-loops may represent shorter Class II-type transcripts not hybridised to the E6 region. Transcripts that form Class III R-loops have not been previously described for BPV-4 and contain the E4/E5 ORF exon spliced at its 3' end to the LI ORF. In Class IV, transcripts map to the 3' end of the EI exon, artificially truncated by the Bam HI site used for cloning the BPV-4 genome, and are spliced to the 5' end of the exon containing the E4/E5 ORF. Class V transcripts are ambiguously located in the cloned BPV-4 genome, and could be derived from the EI or LI ORF. The former case may represent the remainder of the transcript from Class IV R-loops. The rare Class VI R-loops are due to the L2 ORF spliced at its 3' end to the LI ORF.

Sequence homologies between bovine papillomavirus genomes mapped by a novel low-stringency heteroduplex method.

The bovine papillomaviruses (BPVs) types 1, 2, and 5 cause fibropapillomas whereas BPVs types 3, 4, and 6 cause true papillomas. A novel method of heteroduplex mapping at low stringency of hybridisation has identified the position and relative orientation of distantly related sequences in the genomes of these viruses. The genomes of BPV-1 and BPV-2 are closely related but both show a high degree of sequence divergence from the BPV-5 genome. A 1.25-kb sequence adjacent to the unique BamHI site of the BPV-5 genome hybridised to BPV-1 and to the equivalent region of BPV-2. The hybridising sequence in the BPV-1 genome mapped to the C-terminal region of the E1 open reading frame (ORF) and the N-terminal region of the E2 ORF. The BPV-3, BPV-4, and BPV-6 genomes show moderate homology to each other but minimal homology to the fibropapillomavirus genomes. Low-stringency heteroduplex mapping revealed that overlapping sequences in the BPV-1 E1 and L1 ORFs (or the equivalent regions in BPV-2) hybridised to sequences in BPV-3, BPV-4, and BPV-6. Hybrid regions were less than 1 kb long and were sometimes interrupted by short nonhybridising segments. The hybridising sequences in BPV-3 and BPV-4 are positioned in a way that parallels the spacing of the E1 and L1 ORFs in BPV-1. These data suggest that the bovine fibropapilloma viruses and true papilloma viruses share a similar genomic organization, but have undergone extensive sequence divergence.

A novel bovine papillomavirus (BPV-6) causing true epithelial papillomas of the mammary gland skin: a member of a proposed new BPV subgroup.

A papillomavirus has been isolated from frond epithelial papillomas of the bovine udder. It is clearly distinguishable from all other bovine papillomaviruses (BPVs) based on DNA sequence homology and antigenic properties and is thus characterised as a new entity, designated BPV-6. BPV-6 does not possess the interspecific papillomavirus antigen, its genomic DNA (7.2 kb) is smaller than, and does not show any sequence homology to BPV-1, BPV-2, or BPV-5, whereas it is approximately the same length as BPV-3 or BPV-4, with which it shares some sequence homology. The bovine papillomaviruses have been classified into two subgroups: subgroup A, composed of BPV-1, BPV-2, and BPV-5, all of which induce fibropapillomas, and subgroup B, composed of BPV-3, BPV-4, and BPV-6, all of which cause true epithelial papillomas.

The genomes of bovine papillomaviruses types 3 and 4 are colinear.

The 7.2 kb genomic DNA of bovine papillomavirus type 3 (BPV-3) was molecularly cloned using its unique EcoRI site, and the 7.3 kb genome of BPV-4 was cloned using its single BamHI site. The viral genomes were compared by liquid hybridization, Southern blot hybridization and heteroduplex mapping. Low stringency hybridization conditions revealed that the genomes are colinear but the sequences are extensively mismatched. The relative alignment of the restriction endonuclease maps of the two viral genomes has been determined. It was found that the genomes as linearized for cloning are out of phase by 1.7 kb, so that the single EcoRI site of BPV-3 appears to coincide with the BPV-4 EcoRI site at 0.22 map units. It is concluded that the genomes of BPV-3 and BPV-4, both of which cause true epithelial warts, share the same physical organization but exhibit sequence divergence.

Molecular cloning of bovine papillomavirus genomes and comparison of their sequence homologies by heteroduplex mapping.

The genomic DNAs of bovine papillomavirus (BPV) type 1, type 2 and type 4 were cloned in pAT153. BPV1 and BPV2 genomes were cloned using the single HindIII sites of the vector and virus DNAs, and BPV4 was cloned using the single BamHI sites. The orientation of the recombinant DNAs was established by restriction enzyme digestion, hybridization and heteroduplex analysis. The results showed that: (i) BPV1 and BPV2 DNAs are in register and are broadly homologous throughout most of their length when aligned at their single HindIII sites; (ii) depending on the degree of hybridization stringency used, the two DNAs show one major region and several minor regions of partial homology, mainly residing in the segment of the genomes believed to contain the structural genes; (iii) BPV4 DNA shares no homology with either BPV1 or BPV2 DNA.

A B1 repetitive sequence near the mouse beta-major globin gene.

A sequence which lies 2.8 kb to the 3' side of the BALB/c mouse beta-major globin gene has been identified by its ability to hybridise to a member of the human Alu repetitive sequence family. Nucleotide sequencing revealed a 133-bp region that shows 89% homology to the consensus sequence of the B1 family, the murine equivalent of the Alu family. To the 3' side of this sequence is a 31-bp region, C(A)3(C)2T(C)3G(C)11(A)9, which contains oligo(C) and oligo(A) tracts. The whole 164-bp sequence is flanked by a 16-bp imperfect direct repeat, G(A)4GGAGTCTCATAG. The orientation of the B1 sequence is such that transcription by RNA polymerase III would be expected to occur in the same direction as transcription of the neighbouring beta-major globin gene by RNA polymerase II.

Characterization of Alu family repetitive sequences which flank human beta-type globin genes.

Heteroduplex mapping has determined the size, location, and orientation of three Alu family sequences from the human beta-type globin gene cluster. Two of these sequences have the same orientation. One (231 bp long) is 2 kb to the 5' side of the B gamma-globin gene and the other (222 bp) is l kb 5' to the pseudo-beta-l-globin gene. The third (300 bp), 3-4 kb 3' to the pseudo-beta-l-globin gene, has the opposite orientation. Their orientations relative to five previously characterized Alu sequences from this cluster have been established. One of these, 2.5 kb 5' to the epsilon-globin gene, was shown by Southern blot hybridization to be similar but not identical to other family members, whereas the region separating it from a neighbouring inverted repeat is not widely distributed in the human genome.

Repetitive DNA sequences near three human beta-type globin genes.

Five repetitive DNA sequences, of average length 259 bp, have been identified in the intergenic regions which flank three human beta-tupe globin genes. A pair of inverted repeat sequences, separated by 919 bp, was found 1.0 kb to the 5' side of the epsiln-globin gene. Each contains a homologous Alu I site. Another repetitive sequence, with the same orientation as the inverted repeat sequence closest to the epsilon-globin gene, lies about 2.2 kb to the 5' side of the delta-globin gene. A pair of inverted repeat sequences, with the same relative orientations as the other pair and separated by about 800 bp, was found about 1.5 kb to the 3' side of the beta-globin gene.

Denaturation map of the ColE1-Km plasmid pCR11.

The denaturation map of EcoRI-digested pCR11, a ColE1-Km plasmid, is described. The 2.0 kilobase ColE1-derived segment contains an adenine+thymine rich site in the colicin immunity gene region. In the 7.2 kilobase kanamycin resistance region, the transposon Tn903 consists of an adenine+thymine rich 0.98 kilobase kan gene region flanked by a guanine+cytosine rich 1.09 kilobase inverted duplication.

Structure of recombinant plasmids containing synthetic human foetal globin gene sequences.

In vitro synthesized duplex DNA complementary to human foetal globin messenger RNA was integrated into bacterial plasmids and amplified by transformation of Escherichia coli. Recombinants carrying globin DNA were identified by hybridization of foetal globin messenger RNA to bacterial DNA in situ and by liquid hybridization of purified plasmids to specific globin complementary DNA probes. Heteroduplex mapping revealed either a simple insertion loop at the position of the EcoRI site of the parental plasmid DNA. We provide evidence to suggest that these deletions are the result of a site-specific nicking activity of the EcoRI preparations used in the formation of recombinant plasmids.

Recombinant plasmids containing Xenopus laevis globin structural genes derived from complementary DNA.

Details are presented of the in vitro synthesis of double-stranded DNA complementary to purified Xenopus globin messenger RNA, using a combination of reverse transcriptase, fragment 'A' of E. coli DNA polymerase 1 and S1 endonuclease. After selection of duplex DNA molecules approaching the length of Xenopus globin messenger RNA by sedimentation of the DNA through neutral sucrose gradients, the 3'-OH termini of the synthetic globin gene sequences were extended with short tracts of oligo dGMP using terminal transferase. This material was integrated into oligo dCMP-extended linear pCR1 plasmid DNA and amplified by transfection of E. coli. Plasmids carrying globin sequences were identified by hybridization of 32P-labelled globin mRNA to total cellular DNA in situ, by hybridization of purified plasmids to globin cDNA in solution, by analysis of recombinant DNA on polyacrylamide and agarose gels, and by heteroduplex mapping. The results show that extensive DNA copies of Xenopus globin mRNA have been integrated into recombinant plasmids.