The overall strategy for the screen is illustrated in Figure 1. The principle behind the screen is selection against retrovirally infected ES cells. The retrovirus used in the screen carried the puro-Δtk positive/negative selection marker 11. Infected cells expressing the puro-Δtk fusion gene are sensitive to 1-(-2-deoxy-2-fluoro-1-β-D-arabino-furanosyl)-5-iodouracil (FIAU) negative selection; thus, mutant ES cells that cannot be infected by the virus will survive this negative selection. This screen is therefore strongly dependent on the ability to infect all cells in the culture with a retrovirus that reliably expresses a negative selection cassette. A very high infection efficiency must be achieved so that every single infectable cell in the culture has at least one infection event. In practice, the need to infect every cell in a culture of 10⁹ cells requires superinfection, in which every cell has between 10 and 20 independent viral insertions. Superinfection can be achieved in a variety of ways, but in the screen described here it was accomplished by co-cultivation of the gene-trap ES cell library with the viral producer cells.

To generate a viral producer cell line with a high titre, six different murine leukemia virus (MuLV) backbones were tested. The puro-Δtk cassette was cloned into each backbone and the vectors were tested for their efficiency in producing recombinant virus by transient transfection into phoenix packaging cells 12. Viral titers were assessed using wild type ES cells, and the WWF6 (Wang Wei female 6) vector had the highest titer (Figure 2) because of several point mutations and a deletion in its long terminal repeat that prevents transcriptional suppression in ES cells 13. This recombinant vector was used to generate the stable viral producer cell line B4-5 in GPE-86 cells, another safe helper-free ecotropic packaging cell line 14, which produced a titre of 2 × 10⁴ colony-forming units/ml.

Infection efficiency was investigated using both wild-type (AB2.2) and Blm-deficient (NGG5.3) ES cells using irradiated B4-5 producer cells. Following co-cultivation, 99.9% of the both NGG5.3 and AB2.2 ES cells were resistant to puromycin (Figure 3a). Furthermore, proviral copy numbers were investigated using clones isolated from these cultures without selection. Southern blot analysis was performed using an enzyme that cuts once in the provirus, and so each insertion site will have a unique proviral-host genome junction fragment (Figure 3b). This analysis revealed that, irrespective of genotype, all clones had multiple insertion events, with an average of five and a range from two to eight in this analysis. These comparisons confirmed that co-cultivation resulted in very efficient superinfection of Blm-deficient ES cells.

The gene-trap library used in this study is sectored into eight pools, each containing approximately 1,200 independent gene-trap clones recovered by G418 selection (Figure 4a). Each clone in the library has been expanded through a minimum of 14 doublings 9. Given the total cell number, the number of cell divisions and the rate of mitotic recombination in Blm-deficient cells 8, each pool should contain a small number of homozygous mutant cells derived from each independent heterozygous insertion event.

The eight pools were separately co-cultivated with viral producer cells and then selected in FIAU. After the first round of FIAU selection, the surviving FIAU-resistant clones were challenged a second time with the puro-Δtk retrovirus to identify noninfectable mutant ES cells from those that were not infected by chance. After three rounds of selection, 178 infection-resistant clones were recovered from five of the eight pools. At the third round of selection, Blm-deficient ES cells (NGG 5.3) were included as a positive control because they are fully susceptible to viral infection. The phenotype of the infection-resistant clones was clearly different from that of the NGG5.3 cells (data not shown). Two representative clones from each of the five pools were randomly selected to test their degree of resistance to infection (Figure 4b). Among the mutant clones from the same pool the resistance level was quite similar, although it was variable between clones isolated from different pools. Clones from pools 1 and 7 were almost fully resistant to viral infection, whereas mutant clones from pools 2, 3, and 4 were partially resistant, although still obviously different from wild-type cells.

To investigate the relationship between clones in same pools, 23 infection-resistant clones from pool 1 and 27 from pool 7 were analyzed by Southern blotting to detect the proviral-host junction fragment of the gene-trap vector using a SAβgeo probe. This analysis identified the same junction fragment in clones from the same pool (data not shown), implying that clones from the same pool were daughter clones. To determine whether this was the case for clones from all pools, a representative sample of four clones from each pool was tested (Figure 5a). In all cases, clones from the same pool exhibited the same junction fragment, verifying the above assumption. However, clones from the different pools had different junction fragments, confirming that they had independent mutations. A total of five independent clones were recovered, corresponding to five pools of the library.

The proviral-host junction from the five mutant clones was recovered using a splinkerette polymerase chain reaction (PCR) 15 and sequenced. The five sequences mapped to the mCat-1 gene, a known receptor of MuLV in fibroblasts 16. The proviral-host junction sequences mapped to five different positions in introns 1 and 2 of the mCat-1 gene, demonstrating the independent origin of each of these five clones. The mutant clones from pools 1, 7, 2, 3 and 4 were termed V5, V4, V3, V2 and V1, respectively (Figure 5b). In all cases the SAβ geo cassette was in the appropriate transcriptional orientation. Because the ATG initiation codon of mCat-1 is in exon 3, translation of the βgeo fusion gene uses its own start codon. Furthermore, the integration sites of fully resistant clones V5 and V4 from pools 1 and 7 were downstream of exon 2 and closer to the ATG start condon, whereas the partially resistant clones were upstream of exon 2.

To determine whether the mutations were homozygous, these clones were purified by single cell subcloning and Southern analysis was performed using a probe from the mCat-1 gene. This identified different proviral/mCat-1 junction fragments in each clone, as expected (Figure 5c). Moreover, four of the mutant clones (V1, V2, V4 and V5) lacked the wild-type allele and were homozygous for the mutant allele, whereas the fifth mutant clone (V3) appeared to be heterozygous because it retained its wild-type mCat-1 allele.

To confirm the effect of the proviral insertions on expression of the mCat-1 locus, reverse transcription (RT)-PCR was performed using primers for exons 1 and 3, which spanned the proviral insertion sites (Figure 5d). In all cases (including the V3 clone), no product was detected. This confirmed that the viral insertions affected the generation of a wild-type mCat-1 transcript upstream of the normal start codon of mCat-1. Furthermore, the lack of a 'wild-type' exon 1 to 3 RT-PCR product in the V3 clone suggested that this clone may carry a mutation on the 'wild-type' allele that was not identified by Southern analysis. RT-PCR was performed to detect possible transcripts 3' of the proviral insertions. Using exon 4 to 7 and 8 to 12 primer pairs, an aberrant sized fragment was identified in the exon 4 to 7 RT-PCR product from clone V3 (Figure 5d), suggesting a potential splicing mutation on the 'wild-type' allele from this clone. Sequence analysis of this product revealed that the transcript of exon 4 to 7 from clone V3 was shortened by skipping of exon 6 and part of exon 7 (Figure 5d), providing an explanation for the viral resistance of this clone.

Southern blotting analysis using a SAβgeo probe confirmed that each of the five gene-trap clones had only a single gene-trap viral insertion (Figure 5a), although in four out of five cases this was bi-alleleic. To verify that the gene-trap insertions in mCat-1 were directly responsible for the observed phenotype, these were reverted with Cre, which deletes the provirus, leaving a single long terminal repeat in the locus. Reverted clones (1.1R, 7.1R, 2.1R, 3.1R, and 4.1R) were identified following transient expression of Cre by G418-sib selection and confirmed by genomic PCR, with primers for the βgeo cassette (Figure 6a). Excision of the proviral insertion restored the retroviral infection sensitivity of the revertants to wild-type levels for each of the five clones tested (Figure 6b).

To generate a retroviral vector with the highest possible titre, the puro-Δtk positive/negative selection cassette from pYTC37 11 was cloned into several different retroviral vector backbones. The backbones used were pCMV-Babe-Oligo-Revertible (pCBaOR), pBabe-EGFP-Revertible (pBaER), and pBabe-Oligo-Revertible (pBaOR); these three vectors were modified on pBabe retroviral vectors 2526, pQCXIX and pMSCV-Neo (retroviral expression vectors; Clontech, Mountain View, CA, USA) 2728, and pRetro-Super (pRS; a gift from Roderick Beijersbergen, The Netherlands Cancer Institute, Amsterdam, The Netherlands.)

ES cell culture was described in detail previously 30. Briefly, ES cells were maintained on γ-irradiated feeder cell layers in Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum, 2 mmol/l L-glutamine, 50 units/ml penicillin, 40 μg/ml streptomycin, and 100 μmol/l β-mercaptoethanol. Cells were cultured at 37°C with 5% carbon dioxide.

The titer of each retroviral vector was assessed by transient transfection of 25 μg of each vector into phoenix helper-free packaging cells 12 using calcium phosphate transfection 31. Packaged virus was harvested 48 hours after transfection.

The viral titer was assessed using ES cells. Twenty four hours before infection, ES cells were plated in 24-well plates at a density of 3 × 10⁵ cells per well. Viral supernatant from each vector was filtered through a 0.45 μm filter, and polybrene (hexadimethrine bromide) was added to a final concentration of 10 μg/ml. One milliliter of each filtered supernatant was applied to ES cells, and puromycin selection (3 μg/ml) was initiated 24 hours after infection and continued for 8 days. Drug-resistant ES colonies were fixed and stained with 2% methylene blue in 70% ethanol and counted.

pWWF6 (25 μg) was transfected into 1 × 10⁷ GPE-86 cells 14 by electroporation (290 V/cm and 960 μF). The cells were plated and puromycin was added 48 hours after electroporation. Puromycin-resistant colonies were picked into 24-well plates and the titres of 72 independent colonies were assessed as described previously. The 10 clones with the highest titers were reassessed to identify one for use in future experiments. The clone with highest titer was B4-5.

The screen relies on superinfection of the subpools of the gene-trap library. To maximize exposure of the ES cells to viral particles, a co-cultivation strategy was used. B4-5 cells were collected, suspended in media at a density of 1 × 10⁷ cells/ml, and γ-irradiated with the dose of 6000 cGray. About 6 × 10⁷ irradiated B4-5 cells were plated on to a 150 mm tissue culture dish and 24 hours later 3.5 × 10⁶ ES cells from each pool of the gene-trap library were plated onto the irradiated B4-5 cells. After six days of co-culture, the ES cells were confluent. These cells were expanded and 1.8 × 10⁷ cells were selected in FIAU (0.2 μmol/l). Selection was maintained for 10 days. FIAU-resistant ES cell colonies were picked into 96-well feeder plates for further assessment.

Second and third round infection was performed respectively in 96-well and 24-well feeder plates using a sib-selection strategy. For the second round of infection, each clone was plated in duplicate 24 hours before infection and one duplicate was exposed to 200 μl (approximately 4 × 10³ colony-forming units) of viral supernatant from B4-5 cells. This was repeated 12 hours later. The following day one plate was selected with puromycin and a duplicate copy was maintained without selection. Clones that were infected were resistant to puromycin and excluded. Puromycin-sensitive clones were then tested a third time in 24-well plates using the same protocol to confirm their resistance to infection.

Proviral junction fragments were isolated using Splinkerette-PCR, as described elsewhere 15. Briefly, genomic DNA was digested with Sau3AI or FatI and ligated with the corresponding Splikerette adaptors HMSp-Sau3AI or HMSp-FatI. The Splikerette adaptors were generated by annealing Splikerette oligoes HMSpBb-Sau3AI or HMSpBb-FatI with HMSpAa. The first found of PCR was carried out with viral primer AB949new and Splinkerette primer HMSp1. The nested PCR was carried out with the viral primer HM001 and Splinkerette primer HMSp2. The specific PCR products were gel purified and TA cloned 32 for sequencing.

Total genomic DNA was restricted, size fractionated on agarose gels, blotted, and hybridized using standard procedures. The following probes were used: LacZ, a 800 base pair ClaI-digested fragment from pSAgeo 33, which is a plasmid containing the SAβgeo cassette in pBS; and Puro-Δtk, a 1.2 kilobase PstI-digested fragment from pYTC37 11, which is the plasmid that contains the puro-Δtk cassette in pPGKbpA.

Total RNA was extracted from the five mutant clones and NGG5.3 cells as a negative control, and 5 μg total RNA was used to generate the first strand cDNA. β-Actin (Actb) was used as a positive control for RT-PCR. The PCR product of the mutant mCat-1 transcript from exon 4 to 7 from clone V3 was TA cloned for sequencing.

Five million ES cells were electroporated with 10 μg of the Cre expression plasmid pCAG-Cre 34, plated at low density (2,000 cells/90 mm plate), and grown without selection for 9 days. ES cell colonies were picked into 96-well plates. Clones that had excised both copies of the gene-trap cassette were identified as G418-sensitive clones by sib-selection and confirmed by genomic PCR, using primers that amplify a 578 base pair fragment in the βgeo cassette. Reinfection was performed as described above to assess phenotypic reversal.