Isolation of temperature-sensitive diphtheria toxins in yeast and their effects on Drosophila cells

Laser-induced cell ablation experiments have proven to be useful in the study of cell-cell interactions in developing and adult organisms (Sulston, 1988). They allow us to examine the fate of specific cells, to determine cell lineages and to assess the role of cells in vivo. Unfortunately, ablation experiments are often technically difficult to perform and many cells of higher eukaryotes are not easily accessible. Therefore, Palmiter et al. (1987) initiated cell-lineage-specific ablation experiments in mice by introducing the gene for the fully active “A” subunit of the diphtheria toxin (DT-A) under the control of cell-specific regulatory sequences. This elegant approach to studying specific cell lineages has also been employed to define the role of specific cells in living mice (Behringer et al., 1988; Breitman et al., 1987; for review see Evans, 1989), and Drosophila (Kunes and Steller, 1991). However, the system is not generally applicable. First, constitutive expression of the toxin in essential cells is lethal, and precludes the establishment of stable lines for further study unless the toxins can be expressed conditionally (Kunes and Steller, 1991). Second, relatively few highly time- and tissue-specific regulatory sequences of genes have been characterized. Such sequences may be difficult to identify as many genes that seem to be expressed in specific tissues at a particular time of development are often differentially expressed in other tissues or at other developmental stages (Bellen et al., 1989; Wilson et al., 1989; Grossniklaus et al., 1989). These regulatory sequences cannot be used for ablation experiments, although they are tissue specific at some stage of development. Third, temporal control of the expression of the toxin in a particular tissue or cell is dependent on the regulatory sequences that control the expression of the toxin. This temporal control cannot be easily manipulated. Thus, there are almost no cases in which one can govern time, duration and tissue-specificity of gene expression. To circumvent these problems, we have isolated temperature-sensitive variants of the diphtheria toxin in yeast.

Diphtheria toxin was chosen because its mechanism of action is understood (Collier, 1975), its sequence is known (Greenfield et al., 1983), its expression has been adapted for eukaryotic cells (Maxwell et al., 1987), and it is active in all eukaryotic cells tested (Pappenheimer et al., 1980) including yeast (Perentesis et al., 1988). The toxin’s action is mediated by the inhibition of protein synthesis through ADP-ribosylation of elongation factor 2 (for review see Pappenheimer, 1977). In addition, recent but controversial evidence indicates that the toxin also acts as an endonuclease in vitro (Chang et al., 1989; Bruce et al., 1990; Nakamura and Wisnieski, 1990). A single molecule of the fully active toxin is sufficient to kill a mammalian cell (Yamaizumi et al., 1978). The toxin has two subunits. The B subunit is required for receptor binding, whereas the A subunit used in this study is the toxic catalytic domain. Expression of DT-A inside cells usually leads to death without apparent damage to neighboring cells. Should any DT-A escape (e.g from death cells) the absence of DT-B precludes cell attachment and acid translocation.

pSE979 was constructed by means of a three-way ligation of an EcoR1-Smo1 GALI fragment derived from pSE278 (unpublished data), an EcoRV-Voil DT-A fragment from pSE968 (unpublished data, this DT-A fragment was originally present in p2249-l, Maxwell et al., 1987) into the EcoR1-Not1 sites of pUN35 (Elledge and Davis, 1988). The restriction map of this and other vectors is shown in Fig. 1.

The pRXC4 (where X=M, R, I and N ) plasmids were constructed using three-way ligations of the following fragments: a BamH1-Kpn1 fragment from the ninaE promoter sequences from plasmid pDPS48 (a kind gift from Dean P. Smith and Charles Zuker) and the Kpn1-Xba1 DT-A sequences (fully active (=D)) or temperature-sensitive toxins (=M, R, N, I) from pSE979 (D) or mutagenized pSE979 (M, R, I, N) into the BamHl-Xbal sites of pCaSpeR4 (Pirrotta, 1988). Only the most relevant restriction enzyme sites are shown in Fig. 1. The restriction enzyme sites that are shown in bold can be used to replace the ninaE promoter with regulatory sequences of other genes.

To sequence the wild-type and mutant DT-A^ts clones, we subcloned the Kpn1-Xba1 fragments referred to in Cloning Methods into Bluescript KS and SK. Single-stranded DNA was sequenced by the dideoxy chain-termination method (Sanger et al., 1977) with the automated fluorescence procedure (Applied Biosystems) as described by Smith et al. (1986). This allowed us to obtain one full-length sequence in all cases. To obtain the complete sequence on both strands, we used four internal primers to the DT-A. These additional sequences were obtained using the manual dideoxy chaintermination method on double-stranded DNA using ³⁵S as label.

y w fly strains were obtained from V. Pirrotta. Fly strains used to map the various insertions were obtained from Ed Grell. The yw; Ki p^pP[delta 2-3] strain was obtained from Y. Hiromi.

Flies were cultured on standard fly food supplemented with dry yeast at 16°C, 18°C, 25°C or 28°C in a water bath or a fly incubator. No substantial differences were observed between the two treatments. Unless stated otherwise, flies are grown for their entire life cycle at a specific temperature. Temperature shifts prior to pupariation (up or down) do not have any obvious effect on the morphology of the photoreceptor cells as long as the pupae and adults are kept at 25°C or 28°C. Temperature shifts are performed by transferring a vial containing mid- to late-stage pupae or 0-12 hour freshly emerged flies from an incubator to a water bath set at 16°C or 18°C for various lengths of time. The vials were subsequently returned to an incubator at 25°C or 28°C for 1-4 days.

Flies were transformed by injecting 300 ng/μl of the appropriate P-element vector (derivatives of pCaSpeR 4) into yw embryos according to standard protocols using the wings clipped vector as a helper (Rubin and Spradling, 1982). P-element insertions were assigned to chromosomes with standard crosses using y w; CyO/Pm^88k and y w; TM6B, D³/Ki H^B55 stocks, and many P-element insertions were cytologically mapped using digoxygenin-labeled probes.

We injected about four times as many embryos with the P[RDC4] constructs than for the other constructs. Two male P[RDC4] transformants were sterile but two females were fertile. Both original transformant strains established from these females are male sterile. In order to obtain additional transformants, we created transposant strains using the P[delta 2-3] methodology (Cooley et al., 1988; Robertson et ah, 1988). A total of 9 fly strains carrying the P[RDC4] construct were established. Some of these fly strains are homozygous viable and male and female fertile. We generated at least five independent transformants for each construct.

To create mosaic eyes, flies carrying P[RDC4] were crossed to y w; Ki p^pP[delta 2-3]. Progeny carrying both P[RDC4] and the Ki p^pP[delta 2-3] chromosome and exhibiting patches of red and unpigmented ommatidia were analyzed with the compound scope. Some heads were prepared for transmission electron microscopy (TEM) studies.

Single yeast colonies were grown in glucose media lacking tryptophan to select for pSE979 for two days. 10 μl of this culture was plated out on galactose media lacking tryptophan. The plates were inspected daily and the colony sizes were compared with those of cells containing the positive control (pUN35). The data shown in Fig. 2 correspond to a visual estimate of the size of the colonies on a scale of 6 at day three for plates kept at 37°C, 30°C, 28°C and 25°C (0=no growth, 6=size of yeast cell colonies carrying the pUN35 plasmid). The data reported for cells grown at 18°C were scored on day 6.

Heads of flies were fixed in 100 mM sodium phosphate (pH 7.0), 5 mM EDTA, 3% formaldehyde and 2.5% glutaraldehyde. After washing the heads with ddH₂O, the heads were fixed in 2% osmium tetroxide. Heads were embedded in historesin or Spun’s and semi-thin sections (3 μm) and thin sections (0.08 μm) were prepared for light microscopy and TEM.

To isolate temperature-sensitive DT-A (DT-A^tsX), DTA (Maxwell et al., 1987) was placed under the control of the GALI promoter on a yeast centromeric vector, pUN35, that contained TRP1 as a selectable marker (Elledge and Davis, 1988). This plasmid, pSE979, was used to transform the diploid CRY 3 yeast cells. A diploid strain was chosen because there are five recessive genes in yeast that confer resistance to the toxin when mutant (Perentesis et al., 1988), leading to high background reversion frequencies in haploids. CRY 3 (pSE979) cells grow on glucose-containing medium but fail to form colonies on galactose-containing medium, unlike the CRY 3 (pUN35) cells harboring the vector without the DT-A. CRY 3 (pSE979) cells were mutagenized with ethylmethane sulfonate (EMS, survival rate about 50%), allowed to recover for two generations and plated on media lacking tryptophan and containing galactose as a carbon source at 37°C. Approximately 600 colonies grew at 37°C. 22 of the 600 colonies either failed to grow or grew very poorly at 18°C. It was shown that the temperature sensitivity was plasmid-dependent in at least four cases by retransforming yeast cells with the isolated mutant pSE979 plasmids. As illustrated in Fig. 2 and Fig. 3, the CRY 3 (pUN35) cells grew normally on selective medium containing galactose. In contrast, CRY 3 (pSE979) expressing wild-type DT-A did not grow on this medium at any temperature tested (not shown). CRY 3 expressing DT-A^tsX (X=M,N,I, and R) grew at elevated temperatures (37°C, 30°C and 28°C) but failed to grow or grew poorly at 18°C, but each independent ts-isolate displayed a characteristic temperature-dependence profile. As shown in Fig. 3, yeast cells containing the DT-A^TSM at 18°C form a very small translucent colony that can barely be seen. Yeast cells carrying DT-A^tsX are not always killed; some are only arrested or very severely retarded in their growth at 18°C. When the plated yeast cells expressing the DT-A^{ts M}’ ^{1 and R} toxins are held at 18°C for 7 days and subsequently shifted to 25°C, few will resume growth and form colonies within 2-3 days. However, the cells harboring the DT-A^tsN require an extra day to recover. Hence, we conclude that cellular functions required for the growth of yeast are very severely affected at 18°C, but not at elevated temperatures.

To establish whether the DT-A^ts mutant variants isolated in yeast were also active and temperature sensitive in multicellular non-homeothermic eukaryotic organisms, they were introduced under the control of the ninaE promoter into the fruitfly. The ninaE gene is expressed abundantly in the R1-R6 photoreceptor cells (Mismer and Rubin, 1987) that are present in each of the ommatidia of the compound eye. Expression of the ninaE gene starts during the final stages of photoreceptor cell differentiation in mid pupae (48-60 hours after puparium), and continues in adult flies (Zuker et al. 1985). The gene is not known to be expressed at a detectable level in cells other than retinal cells and the larval photoreceptor organ (Mismer and Rubin, 1987; Pollock and Benzer, 1988). We surmized that it should be possible to recover transformants that express the fully active DT-A as well as transformants that express the DT-A^ts, thereby allowing comparison of the defects caused by the different toxins in vivo. In addition, expression of DT-A in R1-R6 cells offers other advantages: first, transillumination of the head permits identification of the pseudopupil and rapid detection of defects in the organization of the photoreceptor cells and their rhabdomeres (Franceschini and Kirschfeld, 1971); second, the eight individual photoreceptor cells (R1-R8) can be easily identified on the basis of morphological features and position within each ommatidium in semi-thin histological sections; third, DT-A expression in the R1-R6 cells should not affect R7, R8 and other ommatidial cells if the DT-A toxins are cell autonomous.

We cloned the four DT-A^ts (M, N, I and R) genes as well as the wild-type DT-A gene under the control of the ninaE regulatory sequences (Mismer and Rubin, 1987) in the CaSpeR4 transformation vector (Pirotta, 1988) . These plasmids named pRXC4 (R=ninaE promoter, X=DT-A M/N/l/R or D for the fully active toxin, C4=CaSpeR4), were injected into Drosophila embryos. Many transformant flies were recovered that contained P[RMC4], P[RRC4], P[RIC4] or P[RDC4]. No transformant flies were recovered that carried P[RNC4], Flies of all these strains, except those that carry the fully active toxin, fail to show external morphological defects at any temperature. However, flies of the nine independent transformant lines that carry the fully active toxin, have rough eyes and abnormal ommatidia (compare Fig. 4A and B), and often lack bristles. Some strains lack many bristles whereas others lack one or two thoracic bristles only. Males of some of these strains are sterile, or substerile. However, we have been able to isolate and maintain healthy homozygous lines that carry the fully active toxin. Hence, we believe that the bristle defects and male sterility problems are due to ectopic expression of the fully active toxin. Since there is great variability for the bristle defects and fertility between the different strains it is very likely that these defects are the result of position effects (for review see Wilson et al., 1990).

To establish which cells and subcellular components were affected by DT-A, we prepared ultrathin sections for transmission electron microscopy (TEM). As shown in Fig. 4B, the morphology of the R1-R6 cells is severely altered in the P[RDC4] strains. When compared to normal R1-R6 cells (see Fig. 4A), the rhabdomeres of P[RDC4] flies are very small and abnormally shaped, the size of the R1-R6 cells is highly variable, they contain many fewer ribosomes, some cells contain many mitochondria, and the cytoplasm and nuclei stain less densely. In contrast, the R7 or R8 cells maintain essentially normal morphological features (see Fig. 4B), indicating that the DT-A acts in a cell autonomous fashion. Since the compounds used to reveal the cellular structures by TEM stain mostly proteins and nucleic acids and since the rhabdomeres are very small or absent, it is likely that protein synthesis is dramatically reduced or arrested and that the intoxicated cells are non-functional. Other cells and structures of each ommatidium, like cornea and cone cells, seem not to be affected by the expression of DTA. However, many pigment cells are absent or highly abnormal, indicating that DT-A is active in these cells. The partial lack and loss of pigment in many ommatidia can also be seen under the dissection microsope in some homozygous viable P[RDC4] flies. Hence, DT-A is either not cell autonomous for all eye cells or DT-A is expressed in certain pigment cells.

To determine if DT-A is indeed cell autonomous, P-transposase-mediated excisions of P[RDC4] were induced. In many mosaic eyes examined, some clusters of ommatidia were wild-type and white (DT-A and red pigment absent because of the excision of P[RDC4]) whereas adjacent clusters of ommatidia were defective and red (DT-A and white⁺ genes of P[RDC4] expressed). TEM of these mosaic eyes showed that morphologically normal photoreceptor cells or pigment cells can be almost completely surrounded by cells that have the typical morphological defects of DT-A expressing cells (see Fig. 4C for a mosaic ommatidium). These observations indicate that the toxin acts in a cell autonomous fashion, and that the pigment cells express DT-A. Hence, the regulatory sequences of the ninaE gene most likely confer toxin expression to the pigment cells in our constructs. It is therefore possible that the ninaE gene is also expressed at low levels in sensitive pigment cells, and no evidence has been presented to the contrary in previous studies (Mismer and Rubin, 1987; Pollock and Benzer, 1988). Alternatively, the construct causes ectopic expression of the DT-A in pigment cells.

To date, we have established eight strains that carry single copies of the P[RMC4] (=ninaE-DT-A^tsN ) construct. Flies carrying this construct show no roughened eye phenotype, bristle defects or male sterility. Transillumination of the eyes of flies reared at 25°C shows an essentially normal pseudopupil. However, transillumination of the eyes of flies of the same strains reared at 18°C shows an abnormal pseudopupil, indicating that the R1-R6 rhabdomeres of these flies are defective. We observe either no or minor defects in the overall organization and structure of the individual ommatidia or cells at 25°C (see Figs 5A, 6C, E and G). Approximately 5% of the R1-R6 cells in the P[RMC4] transformant strains exhibit an abnormal rhabdomere when the flies are maintained at 25°C. liais indicates that DT-A^TSM activity is strongly attenuated at 25°C but that some residual activity may occasionally cause cell damage. This residual activity is not detrimental to more than 90% of the cells since 10-day-old flies (see Fig. 6C, and E) have the same subtle defects as one-day-old flies (data not shown). In contrast, freshly emerged P[RMC4] flies that are reared at 18°C show severe morphological defects in the R1-R6 cells and the pigment cells, but have seemingly normal R7 and R8 cells (see Fig. 5B, 6D, and F). The morphological defects in R1-R6 cells are less severe than those observed in R1-R6 cells of flies of the same age that carry the fully active toxin (compare Figs 5B and 4B, and Fig. 6B with F and D). The rhabdomeres of the R1-R6 cells are abnormal. Most rhabdomeres are significantly smaller than those of R7 or R8 (normally R7 and R8 rhabdomeres are smaller than those of R1-R6). In addition, many rhabdomeres contain vacuoles that disrupt the organization of the membranes of the rhabdomeres. Finally, as is observed in flies that carry P[RDC4], the pigment cells seem to be affected and many pigment granules are not included in cells, suggesting that the integrity of these cells is affected and that pigment cells may be more sensitive to the toxin than the photoreceptor cells. Hence, we conclude that DT-A¹⁵¹⁴ is temperature-sensitive, attenuated and functional in both yeast and Drosophila cells. In addition, the temperature sensitivity in Drosophila is similar to that in yeast since the DT-A¹⁵¹⁴ toxin shows no overt phenotype in yeast at 25°C but inhibits growth almost completely at 18°C.

To establish the minimal cold treatment necessary to induce obvious morphological defects in the R1-R6 cells of strains expressing DT-A^tsM, we exposed flies to different temperatures for various length of time. The minimal cold treatment that caused obvious morphological defects in differentiated R1-R6 cells was consistently obtained with a 6 hour cold treatment at 16°C of freshly emerged flies (see Fig. 6G, H and legend for description of the defects). Shorter cold treatments (3 hours) led to no obvious defects. In general, cold treatments after photoreceptor differentiation (adults) cause less severe defects than cold treatments during photoreceptor differentiation (mid to late pupae). However, if the cold treatment of adult flies that where raised at 25°C is maintained for 5 to 10 days, severe morphological defects can be induced (see Fig. 6I). Thus, altering the structural features of cells that are fully differentiated appears to be more difficult than altering those of differentiating cells.

Analysis of the six established strains that carry single P[RRC4] (=ninaE-DT-A^tsR) insertions revealed that significant differences exist among various P[RRC4] strains: some strains exhibited obvious defects at 28°C but the defects are much more pronounced at 18°C (compare Fig. 6J and K); others show very subtle or no defects in most ommatidia at 28°C and the defects at 18°C are not very severe (compare Fig. 6L and M). This indicates that DT-A^tsR is attenuated and that its effects are dependent on the position of the insertion in the fly genome. Similar observations were made for the DT-A^tsI although this toxin seems to be more active than DT-A^TSR. Since each P[RRC4] or P[RIC4] strain consistently exhibits a more severe phenotype at 18°C than at 28°C, DT-A^{tsR and I} are also temperature sensitive. Furthermore, as noted for DT-A^tsM, the temperature-sensitive profiles of DT-A^{tsR and 1} in yeast and Drosophila are similar.

The four DT-A^ts genes and the wild-type DT-A gene were sequenced to characterize the molecular alterations underlying the temperature sensitivity (see Fig. 7). DT-A^tsM carries a point mutation resulting in a Gly⁵³ ⟶ Glu⁵³ substitution that has already been reported (Giannini et al., 1984). The corresponding mutant form of diphtheria toxin is known as CRM197.

The three other ts-forms of DT-A that we have isolated have not previously been reported or engineered (Collier, 1990).

In this work, we describe the isolation of four temperature-sensitive diphtheria toxins (DT-A^ts) in yeast that are active at 18°C and have reduced activity or are inactive at 25°C to 37°C. Three of these DT-A^ts as well as the fully active DT-A were transformed under the control of the regulatory sequences of the Drosophila ninaE gene into the fruitfly. This is to our knowledge the first report in which viable and healthy transformants in Drosophila have been obtained that carry fully active or attenuated toxins (for previous attempts see Kunes and Steller, 1991). The activities of DT-A^tsM and ^R are clearly attenuated at all temperatures, with DT-A^tsR activity being most dependent on the transgene position in the Drosophila genome. The activity of DT-A^tsI is the least attenuated as several strains show very severe defects in the R1-R6 at 18°C (not shown). The DT-A^tsM toxin may be the most useful toxin as it causes very few or no defects at 25°C and 28°C (data not shown), and seems to act relatively fast at 16°C since a 6 hour cold treatment has an effect on the morphology of the photoreceptor cells in which it is expressed.

Preliminary experiments show that the DT-A^{taM and R} can easily be integrated in the fly genome under the control of other regulatory sequences such as the yeast regulatory sequences (UAS) that respond only to the yeast transcriptional activator GAL4. Hence, these toxins may become useful tools to express the DT-A^ts in tissues other than the photoreceptor cells using a binary system (Omitz et al., 1991) with on the one hand, enhancer detector strains (Bellen et al., 1989; Wilson et al., 1989) that express the GAL4 gene instead of the β-galactosidase gene, and, on the other hand, the strains that carry the UAS-DT-A^ts. A. Brand and N. Perrimon have developed such a binary system in Drosophila (personal communication.).

Analysis of the sequence of the mutant toxins has revealed a number of important observations that should help resolve controversial issues with respect to the action of the DT-A itself. Indeed, DT-A^tsM was found to carry the same point mutation as CRM197. CRM197 has no cytotoxic effects (Uchida et al., 1973), exhibits structural and biochemical properties that distinguish it from the wild-type toxin (Bruce et al., 1990; Hu and Holmes, 1987; Bigio et al., 1987; Mekada and Uchida, 1985; Papini et al., 1987; Collins and Collier, 1985), does not cause ADP ribosylation of elongation factor-2 in vitro between 14°C and 37°C (B. J. Wisnieski, personal communication), and is much more sensitive to intracellular proteases (Yamaizumi et al., 1982). Interestingly, CRM197 shows high endonuclease activity in vitro (Bruce et al., 1990) between 14°C and 37°C (B. J. Wisnieski, personal communication). It is possible that the endonuclease activity is responsible for its temperature-dependent action. Alternatively, CRM197 is active in vivo due to a thermal refolding, which is not observed in vitro.

In DT-A^tsR’ a C ⟶ T transition introduces a translational stop at amino acid 155. This may produce a readthrough and/or a truncated protein that is attenuated and temperature sensitive. DT-A^tsN contains a G ⟶ A transition in the initiation codon. Peabody (1989) has shown that translation initiation can occur at non-AUG triplets, like AUA, but in that case methionine would still be the initiating amino acid. As DT-A^tsN is temperature sensitive in yeast, it is likely that the initiation of translation occurs elsewhere. In DT-A^tsI a stop codon replaces gin³⁷. Such a short peptide is almost certainly inactive. It is more likely that a rare readthrough replaces gin³⁷ with another amino acid to yield a low-abundance temperature-sensitive protein (Firoozan et al., 1991). The DT-A and DT-A^tsX transformed yeast and Drosophila strains should now permit study of the action of these different mutant toxins in living eukaryotic organisms.

In summary, we have isolated and partially characterized the first functional temperature-sensitive forms of diphtheria toxin. These DT-A^ts will aid research on how the structure of the DT-A domain affects toxicity in living eukaryotic cells and shed some new light on the action of CRM197. Because the DT-A^ts are active in very different cell types in many different organisms, they should enable us to arrest or severely impede cell growth and affect morphogenesis in a broad spectrum of organisms, including some plants, provided that suitable expression systems are available. Since the affected cells may maintain contacts with other cells for extended periods of time, such studies may answer different developmental questions than those answered by classical cell ablation experiments. Finally, the use of yeast as a host for the isolation of conditional mutations in heterologous genes may prove to be a general method for creation of conditional alleles of genes from other eukaryotic organisms less amenable to molecular genetic manipulations.

We thank Cahir O’Kane and Kevin Moffat for suggestions and continous exchange of information on their ricin” project, Clive Wilson and Charles Zuker for useful suggestions, lan and Françoise Maxwell, Michael Glode, Dean Smith, Charles Zuker, and Vincent Pirrotta for plasmids, Sandra Kooyer and Nathalie DeSantis for sequencing the DT-A and DT-A^tsM.I,N,R) Donna Turner for TEM microscopy, and B. J. Wisnieski for communicating unpublished results. We also thank Vincent Pirrotta, Richard Behringer, Steve Cohen, Ron Davis, Grant MacGregor, Stephen Lessnick, and B.J. Wisnieski for critical reading of the manuscript. S. E. is supported by grants from the NIH (GM44664) and PEW foundation. H. J. B. is an assistant investigator of the Howard Hughes Medical Institute.