In Caenorhabditis elegans, transgenic lines are typically created by injecting DNA into the hermaphrodite germline to form multicopy extrachromosomal DNA arrays. This technique is a reliable means of expressing transgenes in C. elegans, but its use has limitations. Because extrachromosomal arrays are semistable, only a fraction of the animals in a transgenic extrachromosomal array line are transformed. In addition, because extrachromosomal arrays can contain hundreds of copies of the transforming DNA, transgenes may be overexpressed, misexpressed, or silenced. We have developed an alternative method for C. elegans transformation, using microparticle bombardment, that produces single- and low-copy chromosomal insertions. Using this method, we find that it is possible to create integrated transgenic lines that reproducibly express GFP reporter constructs without the variations in expression level and pattern frequently exhibited by extrachromosomal array lines. In addition, we find that low-copy integrated lines can also be used to express transgenes in the C. elegans germline, where conventional extrachromosomal arrays typically fail to express due to germline silencing.
THE development of techniques that allow exogenous DNA to be introduced into an organism has transformed many diverse areas of experimental biology. Transgenic DNA constructs have been used to rescue mutant genes, express reporter genes, and test the relationship of gene structure and function in vivo. In many cases, transgenic DNA is maintained within an organism extrachromosomally in the form of a plasmid or a multicopy array. Alternatively, transgenic DNA can be integrated into the organism’s genomic DNA either by random insertion or homologous recombination.
In Caenorhabditis elegans, transgene DNA injected into the syncytial cytoplasm of the hermaphrodite germline undergoes intermolecular ligation and recombination to form multicopy extrachromosomal arrays. Association of an extrachromosomal array with a germline nucleus results in formation of a transgenic embryo. Transmission of extrachromosomal arrays from one generation to the next is dependent on array size and can range from 10 to 90% (Mello and Fire 1995); it has been estimated that these extrachromosomal arrays contain at least 80-300 copies of the injected plasmids (Stinchcombet al. 1985; Fire and Waterston 1989; Melloet al. 1991; MacMorriset al. 1994).
Although extrachromosomal arrays created by germline injection have been used successfully to study patterns of gene expression and to identify genes by phenotypic rescue, they have disadvantages that limit their usefulness. Due to the high number of copies of the transgene in an extrachromosomal array, total transgene expression can be elevated relative to that of the corresponding endogenous gene (Fire and Waterston 1989); in addition, expression pattern can vary from animal to animal due to mosaic loss of the extrachromosomal array (Stinchcombet al. 1985). Further complicating matters, the presence of tandemly repeated sequences in an array can trigger gene-silencing mechanisms (Okkemaet al. 1993; MacMorriset al. 1994; Kellyet al. 1997; Hsiehet al. 1999). Transgene silencing is a particular problem in the C. elegans germline, where high-copy-number extrachromosomal arrays are rapidly silenced after a few generations (Kellyet al. 1997; Kelly and Fire 1998; Seydoux and Strome 1999), limiting the ability of researchers to study germline development and function.
One way to avoid problems associated with high-copy extrachromosomal arrays would be to create transgenic lines by direct insertion of transgenes into chromosomes. Unfortunately, the ease with which extrachromosomal arrays are formed and maintained in C. elegans has made it difficult to identify less frequent events such as chromosomal insertion of transforming DNA. One solution to this problem has been to inhibit array formation by including a “poison sequence.” For example, transgenes containing the C. elegans suppressor tRNA gene sup-7 are unable to form high-copy extrachromosomal arrays after injection of transgene DNA into either germline cytoplasm or oocyte nuclei because the sup-7 gene is toxic when present in high copy number (Fire 1986; Melloet al. 1991). Injection of sup-7-containing plasmids directly into oocyte nuclei, however, can be used to create low-copy integrated lines with 1-10 copies of the transforming construct inserted into a chromosome (Fire 1986; Spiethet al. 1988; Fire and Waterston 1989). Low-copy integrated lines have also been obtained by germline injection of high concentrations of oligonucleotides along with the transforming plasmid (Melloet al. 1991). For both of these methods, however, the low frequency of integrated lines obtained, relative to the number of germline injections, has prevented their general use. In a different approach, γ-irradiation has been used to integrate extrachromosomal arrays into C. elegans chromosomes (Mello and Fire 1995), but the high-copy-number DNA in these integrated arrays can still produce elevated levels of transgene expression and gene silencing (Krauseet al. 1994; Hsiehet al. 1999).
The difficulty of making low-copy-number integrated lines in C. elegans, coupled with the mosaicism and silencing observed in extrachromosomal array lines, has restricted the analysis of gene function in C. elegans. Experiments first carried out by A. Rushforth and P. Anderson (personal communication), and more recently by Wilm et al. (1999) and Jackstadt et al. (1999), have shown that microparticle bombardment can be used to create extrachromosomal arrays in C. elegans. In this article, we describe the use of microparticle bombardment to create low-copy integrative transformants in C. elegans. We find that integrated lines created using this approach typically contain only a few copies of the transforming DNA and can be used to express transgenes in both the C. elegans germline and soma.
MATERIALS AND METHODS
Strains: We used the following strains: LGI, dpy-5(e61), dpy-14(e188); LGII, dpy-10 (e128); LGIII, unc-119(ed3), dpy-18(e364), dpy-1(e1), dpy-17(e164); LGIV, dpy-4(e1166), dpy-13(e184); LGV, dpy-11(e224), sma-1(ru3); LGX, dpy-6(e14). AZ188 [unc-119(ed3)III; azEx1(pDP#MM016b; pAZ75)] contains an extrachromosomal array created by germline coinjection of pDP#MM016b and pAZ75 [pAZ75 contains 11 kb of sma-1 genomic DNA cloned into BS(SK+)]. pOK100.03 contains a multimerized myo-2 C subelement enhancer oligo fused to a myo-2 minimal promoter and green fluorescent protein (GFP; Thatcheret al. 1999). OK0023(cuEx16) contains an extrachromosomal array of pOK100.03 and pRF4 rol-6(su1006); OK0039(cuIs2)IV contains an integrated array created by γ-irradiation of OK0023; both strains were kindly provided by P. Okkema and A. Fernandez. DP132(edIs6)IV carries an integrated array of the UNC-119::GFP fusion pDP#MMUGF12. SS599 [bnEx2(pJH4.52 pie-1::GFP::H2B, pRF4 rol-6(su1006); N2 genomic); S. Strome, J. Powers, M. Dunn, K. Reese, G. Seydoux and W. Saxton, personal communication] carries a complex array containing the pJH4.52 plasmid [a GFP::H2B(F54E11.4) fusion that is expressed in the adult germline under control of pie-1 promoter and 3′ untranslated region sequences; G. Seydoux, personal communication]. Strains produced in this study are listed in Table 1. The transformation plasmid pAZ110, used to create integrated lines that express a GFP transgene under control of the sma-1 promoter, was constructed using pPD95.79 (A. Fire, S. Xu, J. Ahnn and G. Seydoux, personal communication). unc-119(ed3) mutants used for bombardment were grown at 20° on 100 mm Opti-gro plates [nemotode growth media (NGM plates; Lewis and Fleming 1995) supplemented with cholesterol, peptone, and Nystatin] seeded with OP50 Escherichia coli.
Bombardment methods: Microparticle bombardment of C. elegans unc-119(ed3) hermaphrodites was carried out using a BioRad Biolistic PDS-1000/HE with 1/42 gap distance, 9 mm macrocarrier to screen distance, 28 inches of Hg vacuum, and 1350 p.s.i. rupture disc. These settings were based on a protocol for microparticle bombardment of Chlamydomonas (L. Mets, personal communication); alternative settings were not extensively tested.
For each bombardment, 1 μl of 1-2-μg/μl plasmid DNA was coupled to 0.6 mg of 1.0-μm microcarrier gold beads, as described in the PDS-1000/HE user’s manual, and bombarded onto a monolayer of ∼10,000 unc-119(ed3) L4 and adult hermaphrodites (a 75-μl pellet) placed on a 20-mm diameter lawn of OP50 on 60-mm NGM plates. Worms were allowed to recover for 0.5 to 2 hr after bombardment and were then transferred onto two 100-mm seeded Opti-gro plates and grown at 24°. Because unc-119 mutants cannot form dauers, they die in the absence of food (Maduro and Pilgrim 1995), making it easy to identify the non-Unc rescued transformants 7-14 days after bombardment. From each plate containing animals rescued for the unc-119 mutation, individual transformed animals were cloned and their F1 progeny scored for presence of unc-119 mutants. Homozygous stable lines were identified by the complete absence of unc-119 mutant progeny over several generations. Heterozygous lines were identified based on the presence of three distinct classes of progeny: heterozygous transformed animals, homozygous untransformed animals, and a third class of sterile or inviable animals. To ensure that each line was the result of an independent transformation event, we retained only one transformed line from each Opti-gro plate.
Mapping integrative transformants: Chromosomal location of integrated DNA in pDP#MM016b and pAZ81 transformed lines was determined using single worm PCR (Williamset al. 1992) to assay linkage between integrated Bluescript vector sequences and marker mutations on each chromosome. PCR primers 5′GGCCCCAGTGCTGCAATGATAC3′ and 5′AA CACTGCGGCCAACTTACTTCTGA3′ were used to assay the presence of Bluescript sequences. For lines transformed with pAZ132 and pAZ119, chromosomal location of the integrated DNA was determined by linkage of GFP expression to marker mutations.
To map autosomal insertions, transformed hermaphrodites were crossed with marker/+ males; F1 progeny were allowed to self-fertilize, and individual F2 progeny homozygous for the marker mutation were scored for presence of Bluescript sequences or GFP expression. If unlinked, 75% of the marker mutation homozygotes should be positive for Bluescript or GFP expression in homozygous transformed lines; 67% of the marker mutation homozygotes should be positive for Bluescript or GFP expression in obligate heterozygous transformed lines. If the transforming DNA is linked to a marker mutation, the frequency of Bluescript- or GFP-positive animals will vary according to the map distance between the integrated DNA and the marker mutation. In this study, we considered a map distance of <25 cM between the marker mutation and the transforming DNA as evidence of linkage.
To test for presence of transforming DNA inserted into the X chromosome, hermaphrodites from homozygous transformed lines were crossed to wild-type males and the resulting F1 males were crossed with dpy-6 hermaphrodites. If the integrated DNA is unlinked to the X chromosome, 50% of the F2 hermaphrodite progeny from this cross should be Bluescript or GFP positive. If the integrated DNA is linked to the X chromosome, all of the F2 hermaphrodite progeny should be Bluescript or GFP positive.
For two transgenic lines in which we initially mapped the integrated DNA to LGIII, we used the unc-119(ed3) mutation present in the background of the bombarded hermaphrodites as a second marker to confirm linkage to LGIII. We crossed transformed hermaphrodites to wild-type males, allowed the F1 cross-progeny hermaphrodites to self, and cloned out non-Unc F2 animals. If the integrated unc-119 rescuing DNA is linked to LGIII, these F2 animals will segregate 1/4 Unc progeny only when a recombination event has occurred between the unc-119(ed3) allele at the endogenous locus and the integration site of the transgenic unc-119 rescuing DNA. Using this strategy, we found that for the heterozygous line AZ69, 19 out of 207 F2 progeny were recombinant, corresponding to a distance of 10 cM between the unc-119 locus and the integrated DNA. For the homozygous line AZ200, 20 out of 284 F2 progeny were recombinant, corresponding to a distance of 7 cM between the unc-119 locus and the integrated DNA. Since the original unc-119 mutation could be re-isolated from these strains, these data also confirmed that unc-119 rescuing activity in these transformants was not the result of gene conversion or reversion of the unc-119(ed3) mutation.
Southern blotting: Southern blots of HindIII or XbaI digested genomic DNA were carried out by standard techniques using the DIG/CPSD system (Boehringer Mannheim, Indianapolis) and hybridized to probes containing either the Bluescript vector sequence (Stratagene, La Jolla, CA) or the 5.7-kb fragment of unc-119 genomic DNA from pDP#MM016b.
Examination of GFP expression patterns: GFP expression in transformed animals was examined using a Zeiss Axioplan microscope equipped with a 100X Plan-APOCHROMAT lens. In the case of extrachromosomal array lines, only animals positive for the cotransformation marker were scored for GFP expression. Presence or absence of variation in the level of GFP expression from animal to animal within a transformed line was determined by comparing GFP expression levels of individual animals within groups of 5-15 animals. Presence or absence of mosaicism in GFP expression patterns was determined by comparing GFP expression patterns of individual animals to the expected pattern of expression for each GFP expression construct.
Integrative transformation in C. elegans using microparticle bombardment: Microparticle bombardment, a technique in which DNA-coated beads are accelerated to high speeds, allowing them to penetrate cells of the target organism, has been used for transformation of plants, animals, and microorganisms (Klein et al. 1987, 1988; Christouet al. 1988; Kindleet al. 1989; Armaleoet al. 1990; Zeleninet al. 1991; Cassidy-Hanleyet al. 1997). We reasoned that microparticle bombardment would have several advantages over germline injection for the creation and detection of chromosomal insertions in C. elegans. First, since each bead can deliver only a small amount of DNA to the C. elegans germline, the probability of creating large extrachromosomal arrays would be decreased. Second, we observed gold beads in both germline and oocyte nuclei of bombarded animals (data not shown), indicating that bombardment can introduce transgenic DNA directly into the nucleus, which may be essential for integrative transformation (Fire 1986). Finally, since a large number of animals are bombarded simultaneously (∼104/bombardment), even rare events such as chromosomal integration could be expected to occur at a detectable frequency.
To identify transformants, we used a selection based on rescue of the unc-119(ed3) mutant phenotype. unc-119 animals are unable to form dauers (Maduro and Pilgrim 1995), an alternative larval stage that normally allows C. elegans to survive for months without food (Cassada and Russell 1975). As a result, unc-119 mutants transformed with plasmids containing unc-119 rescuing DNA (Figure 1) survive and reproduce while the untransformed animals starve and die. Although positive transformants from microparticle bombardment occur at a relatively low frequency in the total population of bombarded animals (∼0.5 × 10-4), this unc-119-based selection permits the surviving non-Unc transformed animals to be easily identified.
Previous work had shown that it was possible to generate low-copy integrated lines by injecting plasmids containing the sup-7 suppressor tRNA gene into oocyte nuclei (Fire 1986; Spiethet al. 1988). In these studies, the primary role of sup-7 was as a cotransformation marker; however, it apparently also acted as a “poison sequence” to suppress formation of extrachromosomal array lines. Due to the difficulty of making transformed lines using this technique, however, this approach is rarely used at present. We reasoned that microparticle bombardment, coupled with the unc-119 selection strategy, might provide an easier method for generating sup-7-containing integrated lines. We bombarded unc-119 hermaphrodites with pAZ81, a plasmid containing both the unc-119 rescuing fragment and the sup-7 suppressor tRNA gene (Figure 1; materials and methods).
From 36 bombardments with pAZ81, we obtained 10 independent transformed lines (Table 2). Five of these transformed lines segregated only non-Unc animals, suggesting that they were homozygous for a pAZ81 integrant. Each of the other 5 lines produced three types of progeny: transformed fertile animals, nontransformed Uncs, and a mixture of sterile animals, inviable larvae, and dead embryos. This segregation pattern suggested that these were heterozygous lines in which animals carrying one copy of a sup-7-containing insertion were able to reproduce, but that in animals homozygous for the insertion, the increased level of sup-7 activity was toxic or lethal. Alternatively, these obligate heterozygous lines may be the result of an insertion into an essential gene or a DNA rearrangement associated with the insertion event (see below). These bombardments did not produce any transformed lines with the characteristic behavior of extrachromosomal arrays, indicating that presence of sup-7 in the transforming plasmid provides a strong selection against the creation and/or maintenance of extrachromosomal arrays, similar to that observed for germline injections of sup-7-containing plasmids (Fire 1986; Spiethet al. 1988; Melloet al. 1991).
Surprisingly, we found that inclusion of the sup-7 suppressor tRNA was not essential for creation of stable lines using microparticle bombardment. From 17 bombardments of unc-119 mutants with the unc-119 rescuing plasmid pDP#MM016b, we isolated 6 independent lines that produced only non-Unc progeny (Table 2). The complete absence of untransformed animals in these transformed lines suggested that these stable lines were homozygous for a chromosomal insertion of the transforming plasmid. An additional 13 independent transformed lines isolated from this set of bombardments segregated both Unc and non-Unc progeny. Although, on the basis of their segregation patterns, these lines appeared to contain extrachromosomal arrays, it is possible that some of these lines contained chromosomal insertions of the transforming DNA, but that animals homozygous for the insertion were unable to reproduce.
Additional experiments have shown that microparticle bombardment can be used to produce stable transgenic lines in a consistent and reproducible manner. From a total of 110 bombardments using five different plasmids containing unc-119 rescuing DNA (Figure 1), we obtained 27 stable homozygous or obligate heterozygous lines, which corresponds to a frequency of ∼0.25 integrants per bombardment (Table 2). These results demonstrate that microparticle bombardment coupled with unc-119 selection is a simple, efficient means of producing stable transformed lines.
Mapping sites of chromosomal integration: To confirm that the stable lines created by microparticle bombardment were the result of insertion of transgenic DNA into a chromosome, we mapped the location of the transforming DNA relative to known marker mutations (materials and methods). For 11 stable lines, created using four different transforming plasmids, we found that each line mapped to a single linkage group (Table 3). In each case, the initial linkage group assignment was subsequently confirmed by recombination with a second marker mutation in the same linkage group. These results unambiguously demonstrate that the stability of these lines is due to integration of transgenic DNA into the C. elegans genome.
In the process of mapping these lines, we observed that integration can affect recombination in the region surrounding the site of integration. For two lines containing GFP-expressing DNA insertions, AZ213(ruIs33/+)V and AZ212(ruIs32)III, we found that there was an unexpectedly low frequency of recombination between the integrated transforming DNA and genetic marker mutations on the same chromosome (Table 3). In the progeny of ruIs33/dpy-11 and ruIs33/sma-1 hermaphrodites, 0 out of 36 animals homozygous for dpy-11 (map position 0.0) were recombinant for ruIs33; 0 out of 73 animals homozygous for sma-1 (map position +3.5) were recombinant for ruIs33. In the progeny of (ruIs32)/dpy-17 and (ruIs32)/dpy-18 hermaphrodites, only 1 out of 65 animals homozgyous for dpy-17 (map position -2.2) and 0 out of 62 animals homozygous for dpy-18 (map position +8.6) were recombinant for ruIs32. We calculated a lowest expected recombination frequency, based on the genetic map distance between each pair of genetic markers (dpy-11 and sma-1, 3.5 map units; dpy-17 and dpy-18, 10.8 map units), and used a chi-square test to determine if the recombination rates we observed were within normal statistical variance. For both AZ213(ruIs33/+)V and AZ212 (ruIs32)III we observed a statistically significant reduction in the recombination frequencies (P < 0.01).
It is likely that the decrease in recombination observed between AZ212 and AZ213 integrated DNAs and marker mutations are the result of DNA rearrangements associated with integration of the transforming DNA. Regions of decreased recombination have been observed for a variety of DNA rearrangements including translocations, inversions, and deficiencies (Rosenbluth and Baillie 1981; McKimet al. 1988; Rosenbluthet al. 1990; Zetka and Rose 1992). Of the 27 stable lines obtained in this study, 7 are obligate heterozygous lines. These heterozygous lines may have resulted from direct insertion of DNA into essential genes or, in the case of lines transformed with pAZ81, from the presence of too many copies of the sup-7 suppressor tRNA gene (Fire 1986). Alternatively, these obligate heterozygous lines may contain DNA rearrangements associated with the integrated DNA that result in homozygous transformed animals that are sterile or inviable.
Integrative transformants typically contain a small number of copies of the transforming DNA: We examined copy number of the transforming DNA in 22 integrated lines produced using microparticle bombardment. Genomic DNA from each line was digested with HindIII and hybridized to a Bluescript-specific probe on Southern blots (materials and methods; Figure 2). Digests of either pDP#MM016b or pAZ81 with HindIII produce a single 8.7-kb band containing Bluescript sequence, while HindIII digests of pAZ119 and pAZ132 produce 5.3-kb and 4.6-kb Bluescript-positive bands, respectively (Figure 1). Digestion of transforming DNA can also produce novel-sized bands as a result of plasmid concatemerization and rearrangement or plasmid breakpoints created during insertion into a chromosome.
Southern blots of genomic DNA from lines created by microparticle bombardment using the transforming plasmids pDP#MM016b, pAZ81, pAZ119, or pAZ132 showed that these integrated lines typically contain only a small number of copies of the transforming DNA. The pDP#MM016b transformed lines AZ200 and AZ205 contained two and three low-intensity DNA bands, respectively, indicating the presence of only a few copies of the plasmid (Figure 2A, lanes 1 and 3); similar patterns containing two or three DNA bands were also found in AZ173, AZ204, and AZ206 (data not shown). Analysis of the 10 lines created by bombardment with the pAZ81 sup-7-containing plasmid indicated that they also contain only a small number of copies of the transforming DNA (Figure 2B, lanes 1-3 and data not shown). Similar results were observed for lines created by transformation with pAZ119 or pAZ132 (Figure 2C, lanes 1 and 2; Figure 2D, lanes 1-4). These data clearly demonstrate that these stable lines do not carry extrachromosomal arrays, which would contain one to two orders of magnitude more copies of the transforming DNA (Stinchcombet al. 1985; Fire and Waterston 1989; Melloet al. 1991; MacMorriset al. 1994). Out of 22 stable lines examined, we observed a high number of copies of the transforming plasmid in only 1 line, AZ199 (Figure 2A, lane 2). Since the transforming DNA in this line mapped to LGV (Table 3), we know that it is integrated into the chromosome. It is possible that the unusually high plasmid copy number in AZ199 is the result of a two-step process in which an extrachromosomal array was formed and then integrated into the chromosome.
In at least two cases, we have obtained transformed lines that contained only a single copy of the transformation plasmid. Southern blots of HindIII-digested DNA from AZ60 and AZ69, which were transformed with the sup-7-containing plasmid pAZ81, each contained a single band distinct in size from the 8.7-kb band observed when pAZ81 is digested with HindIII (Figure 1; Figure 2B, lanes 1 and 2). These data are consistent with the insertion of a single copy of pAZ81: if multiple copies of pAZ81 were present in this line, there would be multiple HindIII sites in the transgenic DNA (Figure 1), which would create a more complex digest pattern. We confirmed this result using Southern blots of Xba I-digested AZ60 and AZ69 DNA. Hybridizing these blots with either Bluescript- or unc-119-specific probes also produced only a single band for each line (data not shown).
In contrast to the small number of copies of transforming DNA in integrated lines produced by microparticle bombardment, we found a much higher level of the transforming DNA in extrachromosomal and integrated arrays produced using germline injection. Southern blot analysis of AZ188, which contains an extrachromosomal array containing the unc-119 gene, and of DP132, which contains an integrated array of an UNC-119::GFP protein fusion (materials and methods), show that these lines both contain many copies of the transforming DNA (Figure 2A, lane 4 and data not shown). Similarly, we found that OK0023, which carries an extrachromosomal array containing the pOK100.03 myo-2 promoter::GFP expression construct, and OK0039, which contains an integrated array derived from OK0023, both contain a high number of copies of the transforming DNA (Figure 2C, lanes 3 and 4).
Stable transgene expression using low-copy integrated lines: In extrachromosomal array lines created by germline injection, the expression pattern of GFP and LacZ reporter constructs can vary from animal to animal due to mosaic loss of the extrachromosomal array and silencing of transgene expression (Stinchcombet al. 1985; Okkemaet al. 1993; Krauseet al. 1994; Kellyet al. 1997; Hsiehet al. 1999). To determine whether stable lines created using microparticle bombardment would have a more consistent pattern of transgene expression, we bombarded unc-119 animals with pDP#MM016b plasmid derivatives containing either sma-1 or myo-2 promoter constructs fused to GFP (Figure 1; materials and methods). In both cases, we were able to isolate stable homozygous lines (Table 2) and found that animals from these lines consistently expressed GFP in the expected patterns (Figure 3A and data not shown).
We compared the GFP expression of these stable integrated lines to transformed lines produced by germline injection of the same GFP expression constructs. We found that the level of GFP expression in transformed animals carrying an extrachromosomal array containing the myo-2 promoter::GFP expression construct varied from animal to animal and that 42% of the transformed animals had mosaic patterns of GFP expression. In contrast, the integrated array OK0039, and the low-copy integrated lines AZ217 and AZ218, which contain the same myo-2 promoter::GFP construct, did not exhibit variation in the level or pattern of GFP expression. Similar results were also obtained in a comparison of extrachromosomal arrays and stable integrated lines containing the sma-1 promoter::GFP fusion (data not shown).
Although the integrated array lines we have examined exhibit a consistent GFP expression pattern, the level of GFP expression does not appear to accurately reflect the number of transgene copies present in the array. Southern blots of the integrated array line OK0039 showed that this line contained a >10-fold increase in the copies of the transforming DNA relative to the integrated bombardment lines AZ217 and AZ218 (Figure 2C, compare lanes 1 and 2 with lane 4). In contrast, the level of GFP expression in OK0039 animals was only ∼2-fold higher than that of AZ218 and 4-fold higher than that of AZ217. Some of the decreased level of GFP expression per transgene copy may be due to partial or rearranged transgenes unable to express GFP. In addition, however, it is likely that the reduced GFP expression per transgene copy in integrated arrays is the result of gene silencing and/or limits on protein synthesis in the cells where it is expressed (MacMorriset al. 1994).
Germline expression of transgenes using low-copy integrated lines: The most dramatic example of gene silencing in C. elegans is seen in the germline, where transgenes in conventional extrachromosomal arrays are not expressed (Kellyet al. 1997; Kelly and Fire 1998; Seydoux and Strome 1999). Germline silencing may be the result of heterochromatic packaging of the repetitive sequences in extrachromosomal arrays. This hypothesis is supported by the requirement for functional mes-2 and mes-6 genes, which encode proteins related to the polycomb group of transcriptional repressors, to maintain germline silencing of extrachromosomal array transgene expression (Kelly and Fire 1998; Seydoux and Strome 1999).
If germline silencing results from the presence of tandemly repeated copies of transgenes in extrachromosomal arrays, we hypothesized that low-copy integrated lines would not be silenced. To test this hypothesis, we bombarded unc-119 animals with pAZ132, a construct containing the unc-119 gene and a GFP::H2B fusion driven by the pie-1 promoter, which directs expression in the adult germline (Figure 1). From 20 bombardments, we obtained five lines that expressed GFP in the germline (Tables 1 and 2). In one line, AZ212, all of the animals expressed GFP (Figure 3B). Two other lines, AZ213 and AZ214, segregated Uncs, GFP-expressing rescued animals, and dead embyos/inviable larvae in a ratio of 1:2:1; these appear to be obligate heterozygous lines. In all three of these lines, we have observed robust GFP expression for >20 generations. In contrast, animals from the other two homozygous lines, AZ210 and AZ211, were unhealthy and lost both GFP expression and unc-119 rescuing activity over several generations. Two additional pAZ132 lines were rescued for unc-119, but failed to express the transgene, possibly due either to germline silencing or disruption of the pie-1::GFP::H2B transgene.
Using Southern blots, we found that the silenced lines AZ210 and AZ211 contained more copies of the PAZ132 transforming plasmid, as determined by complexity of the digest pattern, than AZ212 and AZ213, which continued to express GFP over many generations (Figure 2D, compare lanes 1 and 2 with lanes 3 and 4). This result indicates that the decrease in GFP expression observed for AZ210 and AZ211 is unlikely to be due to a loss of copies of the integrated transgene. We propose instead that in AZ210 and AZ211 the presence of a higher number of copies of the transgene results in silencing of the inserted transgenic DNA, while the lower number of transgene copies in AZ212 and AZ213 does not activate germline silencing.
Germline silencing can be alleviated by creating complex arrays that intersperse genomic and transgene DNA (Kellyet al. 1997). Although both complex arrays and low-copy integrated lines can be used for germline expression, a comparison of the two types of transformed lines suggests that low-copy integrated lines created by microparticle bombardment have several advantages for germline expression of transgenes. In the case of the plasmid pJH4.52, which contains a GFP::H2B gene fusion expressed under control of the pie-1 promoter (materials and methods), it has been difficult to obtain complex array lines that express the GFP transgene in the germline; lines that do express in the germline often lose expression in <5 generations(G. Seydoux, personal communication). In contrast, three out of the five GFP::H2B-expressing lines that we obtained using microparticle bombardment have consistently expressed the GFP::H2B transgene in the germline for >20 generations.
When we compared the complex array line SS599, which has expressed the pie-1 GFP::H2B gene fusion over many generations (S. Strome, personal communication), with the low-copy integrated line AZ212, which expresses the same pie-1 GFP::H2B gene fusion, we observed several striking differences. Maintenance of the SS599 complex array line required growth at 25° and selection of GFP-positive animals in each generation. In this line, 26% of the animals expressed the Rol phenotype associated with the rol-6(su1006) cotransformation marker present in this array. A majority of the animals with a strong roller phenotype expressed the GFP::H2B transgene, although the expression pattern varied in some animals, suggesting that silencing and/or mosaic loss of the transgene was taking place. In contrast, AZ212 could be maintained at all temperatures between 15° and 25°, and 100% of the animals expressed GFP in a consistent pattern. The ease with which integrated lines can be obtained using microparticle bombardment and the stability of germline expression in these lines indicates that this technique is an improvement over currently available methods for germline expression of transgenes in C. elegans.
This study has demonstrated that microparticle bombardment is a simple and efficient technique for generating stable transgenic lines in C. elegans. We have found that a substantial proportion of the transgenic lines generated by microparticle bombardment contain a low number of copies of the transforming DNA integrated into a chromosome, resulting in stable transmission of the transgenic DNA over many generations. A critical factor in the success of this microparticle bombardment transformation strategy is the use of a selectable cotransformation marker to identify rare transformed animals within the population of bombarded animals and their descendants. For the experiments described in this article, we bombarded unc-119 mutants with plasmids containing an unc-119 rescuing fragment and were able to identify transformed animals based on their ability to survive starvation and on their non-Unc phenotype.
In some cases, the unc-119 gene may be an unsuitable cotransformation marker due to interactions between the unc-119 mutant phenotype and the transgenic DNA. In these instances it should be possible to use other selectable markers, such as temperature-sensitive mutants that are sterile or dead at the restrictive temperature. In preliminary experiments, we have found that the dpy-20 gene can also be used to identify stable transformed lines created by microparticle bombardment (S. Kniss and J. Austin, unpublished results), confirming that it is not essential to use unc-119 as the cotransformation marker.
The process by which chromosomal integration occurs in C. elegans has yet to be elucidated. Previous work had shown that creation of integrated lines containing the sup-7 suppressor tRNA occurred only when the transforming DNA was injected directly into oocyte nuclei (Fire 1986). Similarly, the ability to introduce transforming DNA directly into oocyte and germline nuclei by microparticle bombardment may be critical for successful integration of transgenic DNA. We originally predicted that inclusion of sup-7 in the transformation plasmid would be essential for selection of low-copy integrants. We found, however, that we were able to obtain integrants using transforming DNA that did not contain sup-7, indicating this selection was not necessary. It is also clear that using unc-119 as a cotransformation marker does not create a selection for low-copy transformation, because we have also been able to use it as a cotransformation marker in extrachromosomal arrays created both by germline injection and by microparticle bombardment. These results suggest that introduction of DNA into C. elegans by microparticle bombardment inherently favors the creation of low-copy chromosomal insertions.
The presence of nonrecombining DNA in some of our lines, likely due to rearrangements associated with the site of integration (Rosenbluth and Baillie 1981; McKimet al. 1988; Rosenbluthet al. 1990; Zetka and Rose 1992), suggests that integration occurs during the process of repairing double strand breaks. It is not clear whether or not microparticle bombardment plays a direct role in this process, creating double strand breaks or producing cell damage that induces DNA repair activity. In our experiments, the location of integration sites for each of our transforming plasmids appears to occur at random; we identified integration events on four different chromosomes (I, III, IV, and V) and in some cases have identified integration sites at different locations on the same chromosome. It should be noted, however, that 5 of the 11 identified integration sites map to LGIII, which is also the location of the cotransformation marker gene unc-119. Additional experiments will be required to determine if there are favored sites for chromosomal integration and, if so, whether sequences in the transforming plasmid play a role in determining the site of integration for the transforming DNA.
Using microparticle bombardment, we generated lines that express GFP transgenes in reproducibly consistent patterns in somatic tissues. In extrachromosomal array lines, expression patterns can vary from animal to animal due to mosaic loss of the array (Stinchcombet al. 1985). In addition, it has been observed that lines containing extrachromosomal arrays with the same transgenic DNA can vary widely in their level of gene expression relative to gene copy number (MacMorriset al. 1994). These variations in the level of gene expression are likely the result of context-dependent gene silencing, in which expression of tandemly repeated sequences is repressed (Kellyet al. 1997). Mutations in the tam-1 gene result in hyper-silencing of both extrachromosomal and integrated arrays, while endogenous loci and complex arrays, which intersperse genomic and transgene DNA, are able to express (Hsiehet al. 1999). This correlation between transgene copy number and silencing indicates that gene silencing is regulated in a copy-number-dependent manner. We have found that in our low-copy integrated lines that express GFP transgenes, the level and pattern of GFP expression does not vary from animal to animal. We propose that in these lines, the number of transgene copies is insufficient to activate context-dependent gene silencing in the soma. As a consequence, the level of transgene expression in these lines should more accurately reflect gene copy number. Use of low-copy integrated lines should permit expression of transgenes that are toxic when overexpressed as well as a more accurate analysis of protein function in cases where overexpression may alter the localization or regulation of the protein gene product.
We have found that low-copy integrated lines can be used to express transgenes in the C. elegans germline as well as in somatic tissues. Of five lines created using microparticle bombardment that initially expressed a pie-1-GFP expression construct in the adult germline, three lines have continued to express the GFP transgene for >20 generations. In the other two lines, we have observed silencing of both unc-119 rescuing activity and transgene expression. Interestingly, the two silenced lines have only a slightly higher number of copies of the transgene than the lines that have continued to express the pie-1-GFP expression construct. This correlation between transgene copy number and germline expression suggests not only that the germline has a very low threshold for multicopy sequences but also that it is able to discriminate relatively small differences in transgene copy number. This sensitivity of the germline to copy number, due to germline-specific gene silencing mechanisms (Seydoux and Strome 1999), helps to explain why it has been difficult to generate lines that can consistently express transgenes in the germline. The ability to generate stable lines with consistent germline expression by microparticle bombardment represents an improvement over current methods and should increase the ability of researchers to investigate the regulation of germline development and function.
The ability to create low-copy integrated lines will dramatically increase the approaches available for analysis of gene expression and function in C. elegans. Low-copy integrated lines should express at levels close to that of the corresponding endogenous genes, allowing expression patterns and protein function to be determined more precisely. Previous work has shown that integration of transforming DNA via homologous recombination can occur in C. elegans (Brovermanet al. 1993). Although integration by homologous recombination has been only rarely observed in C. elegans, a reliable method for chromosomal integration of DNA may be an important first step toward making homologous recombination a usable tool in this organism. Similarly, the ability to create integrated lines should make it possible to use enhancer traps to identify gene expression patterns, a powerful approach for gene analysis that has previously been hampered by the inability to integrate reporter constructs into the C. elegans genome.
The authors thank Steve Podos, Margaret Morgan, Chip Ferguson, Betsy Goodwin, and the members of the Austin Lab for their comments on the manuscript. Plasmids and strains used in this study were provided by Andy Fire, Peter Okkema, Anthony Fernandez, Susan Strome, and Geraldine Seydoux. We thank David Pilgrim and Morris Maduro for advice on unc-119 and Brian Ackley for suggesting unc-119 as a cotransformation marker. Laurens Mets kindly provided advice and the use of his bombardment apparatus. Many strains used in this study were provided by the Caenorhabditis Genetics Center. Support for this work was provided by American Cancer Society Grant RPG-98-066-01-DDC and the University of Chicago Faculty Research Fund.
Communicating editor: B. J. Meyer
- Received September 6, 2000.
- Accepted November 21, 2000.
- Copyright © 2001 by the Genetics Society of America