Heat shock results in cell cycle delay and synchronisation of mitotic domains in cellularised Drosophila melanogaster embryos

The effects of heat shock upon gene expression were first discovered in Drosophila, with the finding that temperature elevation or treatment with a number of chemicals could bring about a new puffing pattern on polytene chromosomes (Ritossa, 1962) indicative of the induced expression of a small number of genes. Following a shift from 25°C to 37°C, there is a general decrease in ongoing transcription and translation, and transcription is redirected to the synthesis of heat shock mRNAs, the most abundant of which is the mRNA encoding a 70 kDa heat shock protein (hsp70) (Tissières et al., 1974; McKenzie et al., 1975; Spradling et al., 1977; McKenzie and Meselson, 1977; see Pauli et al., 1992 for review).

It has also been known for some time that if Drosophila embryos are heat shocked around blastoderm or gastrulation, this can result in a number of developmental defects resembling known mutant phenotypes and known as phenocopies (Goldschmidt, 1935; Gloor, 1947). Foe and Alberts (1985) have examined the effects of administering heat shock to early Drosophila embryos. The Drosophila embryo is a syncytium for the first 13 divisions which occur at approximately 10 minute intervals, and which consist of rapidly alternating S- and M-phases (Foe and Alberts, 1983). A G2-phase is introduced following cellularisation which occurs in the 14th cycle. Cell divisions then occur within a series of spatially and temporally regulated domains (Foe, 1989). When Foe and Alberts (1985) subjected syncytial embryos to a 37°C heat shock they found that nuclei arrest in interphase, swell and show abnormal condensation of their chromatin. If mid-cycle 14 embryos were subjected to the same treatment, they show no such effect and were thought to be protected from the effects of the heat shock by hsps that can be synthesised from this developmental stage onwards. In cultured Drosophila cells, however, heat shock appears to affect the organisation of the centrosome, and its ability to nucleate microtubules (Debec et al., 1990).

The high levels of heat induced expression from the hsp70 promoter have made it attractive for examining the conditional expression of a large number of cloned genes in transformed flies. Edgar and O’Farrell (1990) showed that when this promoter was used to drive the ectopic expression of the cell cycle regulatory gene string (stg), mitosis no longer occurred in domains, but cells entered mitosis throughout the embryo. Neither this study, nor that of Lehner et al. (1991), who used the heat shock promoter to drive expression of cyclin A, commented upon any adverse effect of heat shock on cell cycle progression. As we are also interested in the use of this same promoter to study the effects of the ectopic expression of other cell cycle genes, we have carried out an investigation of the effects of heat shock administered during cell cycle 14 on morphogenesis and cell division in wild-type Drosophila embryos. In this paper, we report that such a heat shock pulse also has a direct effect upon the cell cycle, disturbing the timing of the divisions in the mitotic domains, and leading to an artificial synchronisation of cells that would otherwise enter mitosis at different times. Taken together, these observations call for caution in the interpretation of data regarding the control of the cell cycle in Drosophila based on the use of the hsp70 promoter.

Embryos were dechorionated in 100% bleach and thoroughly rinsed in distilled water (at room temperature). They were then placed under 0.7% NaCl and viewed using a dissecting microscope. Embryos that were at the particular stage of development required were selected and transferred into 100 l of 0.7% NaCl in a well of a 96-well micro-titre plate, and were given a 30 min heat shock at 37°C in a ‘PREM’ PCR machine (LEP Scientific) in which the rate of temperature change is approximately 1 deg. C per second. Following heat shock the embryos were allowed to recover at 25°C for the times specified in the text. Although we have seen perturbed timing of divisions in all experiments in which embryos were heat shocked, we initially noted some degree of variability in the extent of synchrony depending upon whether the heat shock was given in a 37°C room or in a water bath, and also upon the type of vessel used. This could reflect the previous observations of Lindquist (1980) who reported that the effects of heat shock varied not just with temperature, but with the speed of the temperature increase. The use of a microtitre plate and PCR machine enabled us to standardise the experimental conditions.

Embryos were fixed and prepared for staining as described in Whitfield et al., 1990. DNA was stained using Hoechst 33258 at 1 μg/ml or PDI also at 1 μg/ml. Immunostaining of cyclin A, cyclin B, tubulin and Bx63 antigen was carried out as previously described by Maldonado-Codina and Glover (1992). Observations were made using a Zeiss Standard microscope equipped for epifluorescence, or a Nikon Optiphot attached to the Bio-Rad MRC600 confocal microscope head.

We wished to use the hsp70 promoter to study the effects of the ectopic expression of cell cycle control genes in cellularised Drosophila embryos. We decided first to study the effects of 30 minute heat shocks upon wild-type embryos. These conditions are similar to those used in the experiments shown by Edgar and O’Farrell (1990) studying the ectopic expression of stg. In their experiments, however, heat shock was administered to embryos at timed intervals after egg deposition. As we were concerned to be able to identify the developmental stage as precisely as possible, we dechorionated embryos prior to heat treatment allowing us to select individual embryos staged by their morphological appearance as being in ‘S-phase’, ‘early G2-phase’, or ‘late G2-phase’ of cycle 14 (Fig. 1) (Edgar and O’Farrell, 1990). We have confirmed that the effects we describe can be seen irrespective of whether or not embryos have been dechorionated.

The S-phase of cycle 14 takes place during a 35-45 minute period immediately following mitosis 13 (Edgar and O’Farrell, 1990). Throughout this S-phase period, membranes grow down from the surface of the embryo to surround the nuclei, which become incorporated into individual cells. Embryos that were in the early stages of cellularisation were given a heat shock at 37°C for 30 minutes and then left to recover at 25°C for 0-60 minutes before fixing and staining with Hoechst. All cells were found to be in interphase immediately following this treatment as judged by the organisation of their microtubules (not shown). However, all the nuclei showed abnormally high levels of chromatin condensation. When such embryos were allowed a 30 minute recovery period at 25°C, we found gastrulation movements had begun, and in most cases the pole cells had invaginated. One would normally see divisions in the early mitotic domains at this developmental stage. However, in these embryos all the cells remained in interphase (Fig. 2). Furthermore, their nuclei still exhibited a high degree of chromatin condensation, although this had decreased to some extent during the 30 minute recovery period. We also noted a detrimental effect of heat shock upon the viability of embryos during this period (see legend to Fig. 1).

In order to study the effect of heat shock upon embryos in early G2 of cycle fourteen, i. e. before any domains enter mitosis, we collected embryos just as the cephalic furrow was forming. The recovery of the embryos from a 30 minute heat shock is shown in Fig. 3. Immediately after heat shock (Fig. 3A-C) all the cells were in interphase even though these embryos have reached a stage of morphogenesis at which the early domains should be undergoing mitosis. It appears therefore, that heat shock delays mitosis while allowing continued morphogenesis. We noted, however, that the nuclei all showed an abnormally high degree of chromatin condensation, as is also observed when heat shock is administered during S-phase (see above). Following a 10 minute recovery period the proportion of cells in the embryos that are in mitosis is striking (Fig. 3D-F). The majority of the cells in the cephalic region together with a substantial proportion of cells from the rest of the embryo appeared to be in metaphase and anaphase. These dividing cells correspond to those domains delayed in interphase by the heat shock, together with cells in domains showing their correct temporal sequence of divisions. Consequently, most of the cephalic region and large parts of the body, corresponding to domains 10 and 11, are now all dividing together. The perturbation of the temporal sequence can lead to an even more dramatic synchronisation of cell division. For example, all of the cells visible in the embryo in Fig. 4 are in prometaphase including the amnioserosa (labelled A in Fig. 4A), and domain B in the cephalic region, which would not normally enter M-phase again (labelled B Fig. 4A). Following a 30 minute recovery period from heat shock the embryos looked normal, with all of the cephalic region having divided, and divisions still occurring in the later domains, N and 19. After a 1 hour recovery period, the embryos look normal and cycle 15 appears to take place with the correct timing and patterns in relation to the morphogenetic development of the embryo.

Similar effects are seen if heat shock is administered later in G2, when gastrulation is completed (see Fig. 1) and when the early domains have already initiated mitosis. Such an embryo is shown in Fig. 5. The early domains, that have completed cycle 14, can be seen as areas of increased cell density (indicated by arrowheads). These cells and the surrounding ones that have not undertaken the 14th division are all in interphase as can be seen from the nuclear localisation of the Bx63 antigen, and the staining of cytoplasmic microtubules. This should be contrasted with an untreated embryo at a comparable stage of morphogenesis (Fig. 5D,E). In this embryo, the mitotic domains can be seen as areas in which the RB188 staining is no longer strongly nuclear, but diffuse, and in which microtubules are organised into mitotic spindles.

We have previously described the relative stages of the cell cycle at which cyclins A and B are degraded in the cellularised Drosophila embryo and in cells of the larval central nervous system (Whitfield et al., 1990). Cyclin A is normally high in the cytoplasm of interphase cells, becomes nuclear in prophase cells but is then abruptly degraded, such that neither metaphase nor anaphase cells show staining. Cyclin B protein is usually degraded a little later during metaphase, such that approximately 50% of metaphase cells show cyclin B staining, with no staining seen in anaphase cells.

Immediately after heat shock in early G2, the level of both cyclin A (Fig. 6B) and cyclin B (Fig. 6D) remained high in cells throughout the embryo, reflecting that they are delayed in interphase. These embryos have reached a morphological stage at which divisions should be occurring in the early domains. The control embryos (Fig. 6E-H) clearly show the expected pattern of early divisions. Domains that have just completed the fourteenth division or are in the process of dividing show no staining either of cyclin A or cyclin B. The cells of the later dividing domains remain in interphase and show high cytoplasmic levels of both proteins.

In the almost synchronous mitoses that occur following a ten minute recovery period the pattern of cyclin A destruction does not appear to occur with the normal timing. However, in untreated embryos, we see high levels of cyclin A in less than 10% of cells showing a metaphase configuration of chromosomes, following recovery from heat shock 65% of metaphase cells have high levels of cyclin A (Fig.7, see arrowheads). Once again a striking proportion of cells are undergoing mitosis in this embryo stained to reveal cyclin A. Cyclin B destruction also appears to be delayed in the divisions that follow heat shock. Rather than there being high levels of cyclin B in 50% of metaphase cells, all metaphase cells show high levels (Fig. 7). It is also our impression that cyclin B is not degraded to the extent seen in untreated embryos immunostained using the same antibody.

We are surprised that the effects we describe on wild-type embryos have not been reported previously since the heat shock promoter has been used to drive the ectopic expression of cell cycle regulatory proteins (Edgar and O’Farrell, 1990; Lehner and O’Farrell, 1991). Edgar and O’Farrell (1990) used heat shock to investigate the effects of the ectopic expression of the cdc25 homologue string from the hsp70 promoter. They observed that following the temperature-induced expression of stg throughout the embryo in the G2-phase of cycle 14, mitotic figures could be found in all cells. Furthermore, cyclin A was degraded throughout the embryo during these mitoses, which were immediately followed by the entry of all cells into S-phase. In these studies, Edgar and O’Farrell applied heat shock to embryos carrying the hs-stg construct at a comparable stage to the treatments used in our present study. As we observe considerable synchronisation of mitosis in wild-type embryos following heat shock, it seems that the synchronisation observed by Edgar and O’Farrell is unlikely to be due to the ectopic expression of stg alone. However, in the data they present, the mitoses induced by hs-stg are highly synchronised throughout the embryo, as is cyclin degradation. In our experiments with wild-type embryos, on the other hand, we have observed embryos in which all cells are undergoing mitosis, and others in which the synchronous divisions are restricted to the early domains depending upon the point at which heat shock was given. Furthermore, we do not see synchronous degradation of cyclin, rather this process appears delayed. It is possible that the extent of the effect of heat-shock upon the cell cycle is critically dependent upon the conditions used (see Materials and Methods). Nevertheless, it remains important to determine the extent to which heat shock per se, and the ectopic expression of stg contribute to synchronising cycle 14 mitoses.

Lehner et al. (1991) have found that the ectopic expression of cyclin A from a heat shock promoter following the 16th round of mitoses does not affect the progression through S-phase as judged by the pattern of BrdU incorporation. The exception are cells of the amnioserosa, which unexpectedly showed BrdU incorporation. They attributed this to be a consequence of ectopic cyclin A expression since cells of this tissue normally cease division following S14. However, we find that these G2 amnioserosa cells of wild-type embryos are still able to undertake chromosome condensation following heat shock in cycle 14. Furthermore, we have been able to see BrdU incorporation into some of these cells in wild-type embryos following heat shock administered under conditions comparable to those of Lehner et al. (1991) (data not shown). It is possible that Lehner and colleagues (1991) may not have observed this effect, if it were dependent upon the precise way in which heat shock is administered (see Materials and Methods). Moreover, it looks as though a higher proportion of amnioserosa cells incorporate BrdU in embryos carrying the hs-cyclin A construct, and so we suggest that cyclin A may accentuate this phenomenon, although it is not entirely responsible for it. This interpretation may find support from the observations that amnioserosa cells also appear to undertake DNA synthesis following the heat shock treatment of cycle 14 embryos carrying the hs-stg construct (Edgar and O’Farrell, 1990).

The effects we observe heat shock to have on cell cycle progression are likely to be a consequence of the disruption of several cellular events. The most severe effects that we observe in S-phase could reflect the fact that this is the major period of chromatin assembly and could be a consequence of the known effects of heat shock on chromatin structure brought about by changes in the post-translational modifications of histones. Nuclei could be particularly sensitive to such stress during early embryogenesis as they are still undergoing rapid rounds of DNA replication in which the chromosomes contain multiple replication loops at approximately 30 kb intervals (Blumenthal et al., 1973). In cells that are in G2-phase at the time of heat shock, cell cycle progression may be most affected by the inhibition of ongoing transcription and translation. The delay in entering M may be a specific consequence of the inhibition of transcription of the cdc25 homologue stg, and translation of these transcripts. However, our preliminary experiments indicate that stg RNA does not appear to be degraded as a result of heat shock (not shown), perhaps explaining the rapid recovery that we observe after heat shock. The cell cycle progression of G2 cells already committed to enter mitosis by the activation of p34^cdc2 kinase, may become blocked if the cytoskeleton were to become disrupted, as shown to occur with intermediate filaments by Walter et al. (1990). Structural and functional alterations to the centrosome are also known to occur following heat shock (Debec et al., 1990). Finally, heat shock is known to induce the synthesis of ubiquitin and of some of its conjugating enzymes (Ozkaynak et al., 1987; Finley et al ., 1987). This process has a cell cycle role at least in S. cerevisiae where one of the genes responsible for transferring ubiquitin to an appropriate substrate corresponds to CDC34, a gene required for the G1-S transition (Jentsch et al., 1987; Goebl et al ., 1988). The ubiquitination of proteins frequently targets them for degradation. Indeed, ubiquitination of the mitotic cyclins appears to be responsible for their degradation at the metaphase-anaphase transition (Glotzer et al., 1991). However, our observations show that the degradation of cyclins A and B does not occur prematurely following heat shock of Drosophila embryos in G2 of cycle 14. On the contrary, it appears to be delayed. The persistence of both cyclins in cells undertaking mitosis upon recovery from heat shock again raises the question of whether cyclin degradation is a requirement for this transition (Murray et al., 1989), or is a consequence of it (Whitfield et al., 1990).

The finding that heat shock per se can have a profound effect upon the progression through the mitotic cycle is not new. Indeed Scherbaum and Zeuthen (1954) found that repeated heat shocks could be used to synchronise the cell division cycle in cultures of Tetrahymena. Heat shock has also long been known to have effects in Drosophila development. If embryos are exposed to heat shock during critical periods, a number of defects termed phenocopies are later seen that resemble the effects of mutations (Goldschmidt, 1935; Gloor, 1947). Thus heat shock given around the time of blastoderm formation can result in bithorax phenocopies (Capdevila and Garcia-Bellido, 1978; Santamaria, 1979; Dura and Santamaria, 1983), whereas heat shocks after gastrulation can induce specific patterns of segmentation disruption (Eberlein, 1986). Nevertheless, the heat shock promoter has found extensive use in analysing the effects of ectopic expression of many developmentally regulated genes in Drosophila, and has provided valuable insights into the regulation of development. However, our experiments show the need to exercise great caution in order to fully account for the direct effects of heat shock per se upon cell division, before reaching any conclusion based on the use of the heat shock promoter as a means of directing the expression of either wild type or mutant cell cycle genes.

We thank Hiroyuki Ohkura for his initial observations of synchronous mitoses in heat shocked cycle 14 embryos, and for his continued interest and discussion of our results. We also thank Cayetano Gonzalez and Luke Alphey for their comments on the manuscript. The work was supported by the Cancer Research Campaign.