Visualization of C. elegans transgenic arrays by GFP

Background Targeting the green fluorescent protein (GFP) via the E. coli lac repressor (LacI) to a specific DNA sequence, the lac operator (lacO), allows visualization of chromosomes in yeast and mammalian cells. In principle this method of visualization could be used for genetic mosaic analysis, which requires cell-autonomous markers that can be scored easily and at single cell resolution. The C. elegans lin-3 gene encodes an epidermal growth factor family (EGF) growth factor. lin-3 is expressed in the gonadal anchor cell and acts through LET-23 (transmembrane protein tyrosine kinase and ortholog of EGF receptor) to signal the vulval precursor cells to generate vulval tissue. lin-3 is expressed in the vulval cells later, and recent evidence raises the possibility that lin-3 acts in the vulval cells as a relay signal during vulval induction. It is thus of interest to test the site of action of lin-3 by mosaic analysis. Results We visualized transgenes in living C. elegans by targeting the green fluorescent protein (GFP) via the E. coli lac repressor (LacI) to a specific 256 sequence repeat of the lac operator (lacO) incorporated into transgenes. We engineered animals to express a nuclear-localized GFP-LacI fusion protein. C. elegans cells having a lacO transgene result in nuclear-localized bright spots (i.e., GFP-LacI bound to lacO). Cells with diffuse nuclear fluorescence correspond to unbound nuclear localized GFP-LacI. We detected chromosomes in living animals by chromosomally integrating the array of the lacO repeat sequence and visualizing the integrated transgene with GFP-LacI. This detection system can be applied to determine polyploidy as well as investigating chromosome segregation. To assess the GFP-LacI•lacO system as a marker for mosaic analysis, we conducted genetic mosaic analysis of the epidermal growth factor lin-3, expressed in the anchor cell. We establish that lin-3 acts in the anchor cell to induce vulva development, demonstrating this method's utility in detecting the presence of a transgene. Conclusion The GFP-LacI•lacO transgene detection system works in C. elegans for visualization of chromosomes and extrachromosomal transgenes. It can be used as a marker for genetic mosaic analysis. The lacO repeat sequence as an extrachromosomal array becomes a valuable technique allowing rapid, accurate determination of spontaneous loss of the array, thereby allowing high-resolution mosaic analysis. The lin-3 gene is required in the anchor cell to induce the epidermal vulval precursors cells to undergo vulval development.


Background
The green fluorescent protein (GFP) of the jellyfish Aequorea victoria has been used extensively for observation in vivo of gene expression and cell morphology in C. elegans [1][2][3][4]. GFP has also been targeted to specific subcellular structures by fusing GFP to various proteins. A technique utilizing a chimeric protein of GFP (S65T) and the E. coli lac repressor (LacI) along with lac operator (lacO) makes the visualization of chromosomes possible [5][6][7][8]. This fusion protein has the DNA-binding capability of LacI and the fluorescent properties of GFP. The fusion protein is capable of binding to the lacO, thus localizing GFP expression at the DNA repeat. Such localization allows direct visualization of segregating chromosomes during mitosis.
We have applied the GFP-LacI technique to C. elegans. We show that the GFP-LacI•lacO repeat technique allows visualization of transgenes present as either extrachromosomal arrays or integrated into a chromosome. The integrated version allows visualization of chromosomal segregation and determination of polyploidy.
Visualization of extrachromosomal arrays provides a method to detect transgenes used for mosaic analysis. Genetic mosaics in C. elegans are typically generated by the spontaneous somatic loss of an extrachromosomal transgenic array or a free duplication [9][10][11][12][13]. When the free duplication or extrachromosomal array containing a wild-type cell-autonomous marker gene (often ncl-1; enlarged nucleoli) and a gene of interest is lost from one of the daughter cells during mitosis, it gives rise to a lineage of cells lacking wild-type activity of the marker gene and of the gene of interest. Extra-chromosomal arrays are mitotically unstable, resulting in a complex mosaic pattern, establishing a method of scoring individual cells under Nomarski differential interference contrast microscopy.
The inductive signal for hermaphrodite vulval differentiation is the epidermal growth factor (EGF) like protein LIN-3 [14,15]. lin-3 encodes proteins that have an extracellular domain with one EGF motif, a transmembrane domain and a cytoplasmic domain. In the presence of the gonadal anchor cell (AC), three of the six vulva precursor cells (VPCs) undergo three rounds of mitosis and generate the cells that form the vulva. The VPCs are the posterior daughters (P3.p-P8.p) of six of the twelve Pn cells present at hatching [16,17]. The VPC (P6.p) nearest to the AC will adopt the 1° fate since it receives more signal than its neighbors. P5.p and P7.p cells are induced to adopt the 2°f ate, either directly by LIN-3 [18] or indirectly via the 1°V PC [19]. The VPCs (P3.p, P4.p. and P8.p) further from the AC adopt the 3° fate, which is to generate two nonvulval descendants that fuse with the hyp7 epidermal syncytium. The fates adopted by the VPCs are distinguished in part by the number of progeny they generate. The 1°a nd 2° cell fates generate eight and seven descendants, respectively, which form the mature vulva [15]. Decreased lin-3 activity results in decreased vulval development while overexpression of lin-3 results in increased vulval development. Genetic epistasis tests indicated that lin-3 acts upstream of let-23, sem-5, let-341, let-60, lin-45, mek-2 and mpk-1 during vulval induction [reviewed by [20]]. Based upon lacZ and GFP reporter gene constructs, lin-3 is expressed in the anchor cell at the time of vulval induction [14,[21][22][23], and in the 1° vulval lineage after vulval induction [24]. Recently, Dutt et al. [25] argue based on molecular genetic experiments that lin-3 can act in the VPCs to extend the range of induction. Based upon its structure, expression and genetic properties, the AC has been proposed to secrete LIN-3 protein [14,18]. We have tested by mosaic analysis, using GFP-LacI•lacO, whether lin-3 signal is required solely in the anchor cell for vulval induction.

Visualization of chromosomes
To test whether the GFP-LacI•lacO system could be used to visualize the DNA of extrachromosomal arrays in C. elegans, we engineered sequences that encode a GFP-LacI fusion protein under the control of the heat-shock enhancer/promoter in vector pPD49-78 (hsGFP-LacI). We then microinjected a DNA mixture containing hsGFP-LacI, the lacO repeat and dpy-20 rescuing DNA (pMH86) into the gonad of an adult dpy-20(e1282) hermaphrodite. After a 30-minute heat-shock at 33°C, transformants were found to express nuclear GFP and have intense foci of subnuclear fluorescence, presumably corresponding to the DNA of the extrachromosomal arrays. DAPI co-staining confirmed that the GFP-LacI•lacO system has nuclear expression and association with DNA. We first detected expression in embryos at early gastrulation (~24 cell stage). Larvae and adults express GFP broadly.
DNA molecules injected into the C. elegans gonad syncytium assemble into arrays; extrachromosomal arrays consist of many rearrangements of the DNA injected [26]. The fusion protein bound to the lacO repeat resulting in one to two bright spots per nucleus as well as the unbound fusion protein resulting in nuclear diffuse fluorescence ("haze") ( Figure 1). Mitotic loss of these mixed extrachromosomal arrays in a single founder cell results in a clone of cells lacking the activities of all genes in the array [11,27].
To test the reliability and consistency of the GFP-LacI•lacO as a detection method for transgenes, we examined animals in which every cell has the lacO array and in which most cells express GFP-LacI. Specifically, non-Dpy transformants of dpy-20(e1282); syEx [pMH86 + pPD49-78GFP-LacI + lacO] were X-irradiated to integrate the transgenic array into the genome, yielding integrated transgene syIs44 in the strain PS2442 (Table 1). In this strain, we observed that most nuclei had one or more spots of fluorescence (see Figure 1B). Detection of chromosomes was efficient. For example, we observed two pairs of sister cells in each of the 20 animals, the embryonic sisters F and U and the postembryonic sisters P8.pa and P8.pp. We found that all 80 cells had one or two fluorescent spots. Therefore, we are able to detect a transgene in every cell.
We then examined the behavior of an extrachromosomal transgene carrying lacO in a strain (PS3427) that expressed GFP-LacI from an integrated trasngene (syIs46). To ascertain whether the loss of the array coincides with the progenitor cell and its two daughters, we looked at the nuclei of two sister cells, namely, F and U ( Table 2). Of the 67 animals viewed with the integrated GFP-LacI and the extrachromosomal lacO, 65.6% of the progenitor cells had a localized bright spot (i.e., array present). Of the 65.6% progenitor cells having a localized bright spot, 15.9% pass the array to only one of the daughter cells.

GFP-LacI • lacO as a single cell marker
The ncl-1(e1865) mutation results in enlarged nucleoli of a large number of cell types (the Ncl phenotype). Since ncl-1 acts in a cell autonomous manner, it is useful as a cell lineage marker [10]. The cosmid clone (C33C3) rescues the ncl-1 phenotype and has been used as a cell lineage marker on extrachromosomal arrays [28][29][30]. To test the utility of the GFP-LacI•lacO technique as a single cell marker, we compared it with ncl-1. A DNA mixture composed of the lacO repeat, C33C3, and lin-15B (+) DNA was injected into the gonads of syIs46; ncl-1(e1865); dpy-20 (e1282); lin-15(e1763) animals. The isolated transgenic mosaic animals were then heat shocked (See Methods). We scored the Ncl phenotype in the nucleoli of 22 different cells (m2, m3L, m3VL, m4, m3R, m3VR, m3DR, m3DL, hyp7 dorsal head, hyp7 ventral head, P/I, p3.p, P4.p, P5.p, P6, p, P7.p, P8.p, B, F, U, hyp7 ventral tail, hyp7 anus) per mosaic animal (n = 37; Table 3). The presence of the spot is unequivocal evidence that the trans- Visualization of C. elegans extrachromosomal arrays by GFP gene is present. The existence of Ncl cells with fluorescent spots thus clearly indicates a false negative by Ncl-1. Conversely, non-Ncl cells without a fluorescent spot suggests either the perdurance of the NCL-1 protein (i.e., false positive) or a false negative by GFP-LacI•lacO. Both markers agree in scoring approximately 80%, while false negatives of Ncl-1 (N, +) occur about 16% ( Table 3). The occurrence of false positives of Ncl-1 (W, -) is 2%. However, neither marker is perfect. The Ncl marker is undetectable in intestinal nucleoli and endogenously large nucleoli such as some hyp cells and muscle cells. The GFP-LacI•lacO system has a higher apparent loss rate per cell division (15.9%; Table 3; discussed later) as well as bleaching of GFP.
One explanation for the lack of intense fluorescent spots (false negatives by GFP-LacI•lacO) is that the expression of GFP-LacI is insufficient. We thought this possibility likely since the expression initially depended on the GFP filter combination used, and the light source (200 watt vs. 100 watt HBO burner). To increase the sensitivity, we engineered a GFP-LacI under the control of the ubiquitously expressed dpy-30enhancer/promoter [31]. Another  The combined use of the cell lineage marker ncl-1 and the GFP-LacI•lacO would increase the accuracy and ease of mosaic analysis. In order to use both markers, the double mutant ncl-1; dpy-20 was injected with the DNA mixture (pMH86 (dpy-20(+) + pPD49-78::GFP-LacI + dpy-30::GFP-LacI). Once non-Dpy transformants were isolated, the extrachromosomal array was integrated into the genome by X-ray irradiation to yield strain PS2958 syIs46 [pMH86 + pPD49-78::GFP-LacI + dpy-30::S65T*LacI] II; ncl-1 III; dpy-20 IV. This strain became the basis for our genetic mosaic analysis of lin-3, discussed below.
The comparison between ncl-1 and the GFP-LacI•lacO as single cell markers indicate that GFP-LacI•lacO is comparable in its reliability to ncl-1 as a cell lineage marker for the presence of a transgene ( Table 3). The main advantage of using the fusion protein GFP-LacI•lacO rather than Ncl is that scoring the mutant Ncl phenotype is typically more difficult than scoring cells with the bound GFP-LacI fusion protein. Also, GFP-LacI can be used in cells such as intestinal cells for which ncl-1 is not applicable. We find that the GFP-LacI method (spot or not spot) is much easier than scoring nucleolus size. Such ease of scoring may also be used for more accurate mosaic analysis. The comparison between the GFP-LacI•lacO and ncl-1 as a single cell marker as well as its loss rate per cell division (Tables 2 and 3) confirms that using both the Ncl-1 phenotype and the integrated GFP-LacI together will increase the accuracy of mosaic analysis. An alternative mosaic marker, the nuclear-localized SUR-5-GFP, demonstrates the ease and speed of scoring cells by fluorescence comparative to the ncl-1 marker [30]. This technique allows to rapidly screen with a dissecting microscope for rare mosaic animals, unlike the GFP-LacI•lacO methodology that requires a compound microscope. SUR-5-GFP is an excellent mosaic marker for determining if the gene of interest is either in the AB or P1 lineage but for finer single cell analysis, either ncl-1 or GFP-LacI•lacO is required. Another limitation of SUR-5-GFP is its limited expression pattern. GFP-LacI•lacO is expressed throughout the somatic cell lineage.
The high apparent loss rate of GFP-LacI•lacO, for example 15.9% per cell division in one experiment, could result from two general causes. First, the presence of lacO repeats might lead to decreased stability of the arrays. Second, the presence of the repeats might increase silencing of the transgenes in some manner [32]. We also believe that the ability to detect rapidly the presence of the transgene in individual cells affords a more accurate picture of the degree of mosaicism than gross phenotypic description.

Polyploidy
Our method for ploidy detection was tested by injecting a DNA mixture consisting of a fusion protein under the control of the heat-shock promoter pPD49-78, the lacO repeat, plus rol-6(su1006) dominant into a lin-5(e1348) mutant, which fails to undergo mitosis [33]. The Rol segregants of strain PS2629 confirmed that GFP-LacI +lacO could detect polyploid cells. Multiple dots have been seen when viewing known polyploid cells (i.e., intestines and hypodermal cells; Figure 2). We conclude that the GFP-LacI•lacO method is useful in determining which cells are polyploid. However, the number of spots within the cell cannot unambiguously determine the extent of ploidy. In particular, diploid cells have 1-3 spots while polyploid cells have ≥ 4 spots.
The number of fluorescent spots per nucleus does not correlate precisely with the expected copy number, raising the possibility that there is some synapsis of chromosomes at Assay for polyploidy of nuclei some stages of the cell cycle. Nonetheless, while examination of a single cell is not sufficient, examination of several cells would be, and thus this method could be used for screens or as an assay for alterations in ploidy.
We examined animals with a fluorescent pharynx due to expression of MYO-2::GFP. These animals have the transgene in either the AB lineage or P1 lineage or both [36]. We picked L3-L4 animals expressing myo-2::GFP by viewing them under a dissecting microscope with a GFP filter. These animals were then heat-shocked for 30 minutes in a 33°C water-bath followed by a one-hour recovery period in a 20°C incubator.
We examined a total of 114 animals prescreened under a dissecting stereomicroscope. We considered three possibilities: lin-3 acts in the AC, it acts in the VPCs or both since lacZ and GFP reporter gene constructs indicate lin-3 expression in the anchor cell at the time of vulval induction [14,[21][22][23], and that it acts in the 1° vulval lineage after vulval induction [24,25]. We scored the AC and the VPCs of wild type, Vul and Muv lin-3 transgenic animals.
Of the 114 animals, 91 had wild-type vulva, 15 animals were vulvaless and eight animals were multivulva. Eightyeight animals with the array present in both the AC and VPCs were wild type or Muv. All eight animals lacking the array in both the AC and VPCs were Vul, indicating that lin-3 is necessary either in the anchor cell, the VPCs or both. Of nine animals that had the array in the AC but not in the VPCs, eight were wild type and one was Muv (Table  4, Figure 3A-D), indicating that expression of lin-3 in the anchor cell is sufficient for vulval induction. All seven animals that had the array in the VPCs but not in the AC were Vul (Table 4 and Figure

Conclusion
We demonstrated that GFP-LacI•lacO can be used to visualize transgenes in living C. elegans, either as extrachromo-  somal arrays or as arrays integrated into a chromosome. We also showed GFP-LacI•lacO as a useful marker for ploidy determination in vivo. In addition, we showed that this direct visualization using transgenes facilitate the high-resolution of mosaic analysis. The ease and accuracy of detecting mosaics using GFP-LacI•lacO were examined by comparing its efficiency with ncl-1 as a single cell marker. We demonstrated the utility of GFP-LacI•lacO as a mosaic marker by demonstrating that the lin-3 site of action for vulval induction is in the anchor cell. Overall, GFP-LacI•lacO is a useful tool for several aspects of C. elegans molecular genetics. Indeed, our system have been successfully used in several experiments by others to colocalize proteins binding DNA and for determination of ploidy [37][38][39][40][41][42][43].

Nematode methods
Growth and handling of C. elegans strain N2 were according to Brenner [44]. All experiments were performed at about 20°C unless otherwise stated. The genetic and cellular nomenclature of C. elegans was followed according to [45] and [17], respectively.

GFP-LacI fusion and 256 lacO repeat
The GFP-LacI fusion protein and the lacO repeat [6][7][8] were graciously given to us by Andrew Belmont. We placed GFP-LacI under the transcriptional control of an hsp16 promoter/enhancer element pPD49-78 [47,48] The fusion protein was inserted into KpnI/SacI site of heat shock vector pPD49-78. pPD49-78 is expressed very well in the neural and hypodermal cells, as well as in the gut, muscles, and pharynx but not in the germline [47,48]. The GFP-LacI coding sequences were also placed under the transcriptional control of the dpy-30 promoter graciously provided by Barbara Meyer. The dpy-30 promoter/ enhancer directs expression throughout the animal's somatic cells. [31]. The GFP was replaced with GFP(S65T) from the vector pPD93-65, which contains introns, in order to increase translation of the fusion protein [47].