Genome-wide RNAi screen for genes of temperature-dependent importance

To identify cell proliferation genes with a requirement that depends on growth temperature, we performed a genome-wide RNAi screen with D. melanogaster S2R + cells (Fig. 1a). These adherent cells are well suited for image-based screening. While usually cultured at around 25 °C, the presumed temperature optimum of the organism D. melanogaster, S2R + cells proliferate also at considerably lower temperatures, although with reduced cell spreading and increased aggregation (Bai et al. 2021). At 17 °C, S2R + cells are still fairly spread out. Moreover, at this temperature, an increase in cell numbers is readily detectable at day 10 after re-plating (Bai et al. 2021), and mitotic figures are also present at this time point (Fig. 1a). At 27 °C, a comparable increase in cell numbers required about 3 days according to our initial comparisons. Therefore, to compare the effect of knockdown of a particular gene at either 17 °C or 27 °C, RNAi treatment was done for 10 and 3 days, respectively, before cell analysis. During the screen, the cell number increase resulting at 27 °C was somewhat greater than that at 17 °C (Fig. 1b, compare no dsRNA at 17 °C and at 27 °C). RNAi appeared to be comparably effective in the two chosen conditions, as suggested by Diap1 knockdown. Diap1 codes for an essential anti-apoptotic factor, and its knockdown causes apoptosis (Boutros et al. 2004). The reduction in cell numbers after Diap1 knockdown at either 17 °C for 10 days or 27 °C for 3 days was similar (Fig. 1b). The RNAi library HD2 was used for screening (Horn et al. 2010). This genome-wide library comprises sixty 384-well plates and targets 14,587 D. melanogaster transcripts. Four identical replicates with dsRNA aliquots from the library were generated (Fig. 1b). Two replicates were used for S2R + cell treatment at 17 °C and two at 27 °C. For image-based analysis after dsRNA treatment, we performed double labeling with a DNA stain and a green fluorescent antibody against α-tubulin (Fig. 1b).

Fig. 1

Genome-wide RNAi screen for genes of temperature-dependent importance. (a) Design of the RNAi screen. After exposure of S2R + cells to dsRNA at either 17 °C for 10 days or 27 °C for 3 days, cells were double labeled with a DNA stain (DNA) and green fluorescent anti α-tubulin (Tub) for image-based analysis. The images (bottom) display a region from a control well after incubation at 17 °C without dsRNA (see also dashed white rectangle in panel b) with mitotic figures indicated by arrowheads: metaphase (m) and telophase (t). (b) Representative examples for effects observed in the screen after treatment with the indicated dsRNAs for 10 days at 17 °C (left) or 3 days at 27 °C (right). Scale bars = 20 µm (a) and 100 µm (b)

The final images obtained in the screen are illustrated with a few characteristic examples (Fig. 1b). Wells without any dsRNA (negative controls) contained a relatively high number of cells at both temperatures (Fig. 1b). As already stated above, knockdown of Diap1 resulted in a strong reduction of cell counts at both temperatures (Fig. 1b). Similarly, a reduction in cell numbers at both temperatures although to a lesser extent resulted after knockdown of Cdk1 (Fig. 1b), which is essential for progression through the cell cycle (Stern et al. 1993). In contrast, knockdown of some genes, including ballchen (ball), had an effect that was clearly dependent on temperature. In case of dsRNA targeting ballchen (ball), cell number was more drastically reduced at 17 compared to 27 °C (Fig. 1b).

For an initial evaluation of the technical quality of the screen, we used low magnification images of the DNA signals, which revealed an obvious reproducibility between replicates, for example, for the negative and positive control wells (Fig. 2a). These low magnification images were also used for an initial identification of genes with a requirement that appeared to be temperature-dependent. For this initial screen hit identification, we averaged the images of the two same-temperature replicates. Visual inspection of the resulting average images confirmed the presence of some wells with a reduction of the overall DNA content predominantly at only one of the two analyzed temperatures (Fig. 2a).

Fig. 2

RNAi screen hits with temperature-dependent effects on cell numbers. (a) Overall DNA content in wells revealed by low-magnification images of the DNA signals. The equivalent regions of 384-well plates of all four replicates, as well as an average image of the two same-temperature replicates are displayed. Dashed circles indicate a negative control well without dsRNA (n), a positive control well with Diap1 dsRNA (p), a well with reduced overall DNA content at both temperatures (b), and a well with low DNA predominantly at the low temperature (l). (b, c) Scatter plots (left) display the cell count B-scores (mean B-score of the two same-temperature replicates) after knockdown at either 17 °C or 27 °C for all the tested dsRNA amplicons. Venn diagrams (right) indicate the numbers of dsRNA amplicons classified based on the temperature dependency of their effects on cell counts. The number of genes identified by these dsRNA amplicons is given below the dsRNA amplicon number. (b) Values of dsRNA amplicons resulting in unusually low cell numbers (average B-score ≤  − 2) are colored in case of amplicons associated with reduced cell counts at only 17 °C (blue), at only 27 °C (red), and at both temperatures (green). (c) Values of dsRNA amplicons resulting in unusually high cell numbers (average B-score ≥  + 2) are colored in case of amplicons associated with increased counts at only 17 °C (blue), only 27 °C (red), and at both temperatures (green)

For a more quantitative hit identification, cell counts were determined for each well by automated analysis of high magnification images. B-scores (Brideau et al. 2003) of the obtained cell counts were calculated (Online Resource 1, S1 Table). Analysis of the B-score distribution indicated that the screen data was not distorted by severe technical artefacts affecting certain plates or replicates (Online Resource 2, S1 Figure). The reproducibility of B-scores obtained for the two 27 °C replicates was higher overall than in case of the two 17 °C replicates (Online Resource 2, S1 Figure). Although technical reasons for the lower reproducibility at the low temperature are not excluded, it might also reflect increasing phenotypic decanalization at temperatures further away from the optimum, as proposed in case of transcriptome changes in adult D. melanogaster (Chen et al. 2015). Compared to the overall correlation between the two 27 °C replicates (R2 = 0.63), the correlations between 27 and 17 °C replicates were lower (R2 = 0.44, 0.37, 0.47, and 0.39 for 27_repl1 vs 17_repl1, 27_repl1 vs 17_repl2, 27_repl2 vs 17_repl1, and 27_repl2 vs 17_repl2, respectively). However, the correlations between 27 and 17 °C replicates were not consistently lower in comparison to the overall correlation between the two 17 °C replicates (R2 = 0.39).

Because only two replicates per temperature were analyzed in the genome-wide RNAi screen, further validation of screen hits appeared to be crucial. To identify screen hits that might deserve further validation, we focused on dsRNA amplicons resulting in cell counts that were either unusually low or unusually high (B-score of cell count ≤ -2 or ≥  + 2). About 25% of all analyzed dsRNA amplicons resulted in such unusual cell counts, in case of all four replicates. Unusually low cell counts were observed about two-fold more often than unusually high cell counts, consistent with the notion that genes required for growth, proliferation, or survival of S2R + cells are more numerous than genes with negative effects on these processes, as clearly confirmed for yeast (Yoshikawa et al. 2011).

In a second step, we selected dsRNA amplicons that affected cell counts differentially at the two analyzed temperatures (Fig. 2) and generated several initial hit lists (Online Resource 3, S2 Table).

A first hit list, l17weak, contained genes associated with unusually low cell numbers after knockdown at 17 °C (average B-score ≤ -2), while cell counts were at most slightly reduced at 27 °C. Eight hundred nine genes were on this list, identified by 923 distinct dsRNA amplicons resulting in weak cell proliferation preferentially at 17 °C. A similar list (l17weak_a: 150 genes identified by 162 dsRNA amplicons) was generated from amplicons resulting in a reduction of cell counts at both 17 and 27 °C but far more extensively at the lower temperature. These two lists contained candidate genes more important for cell cycle progression, growth, and survival at low temperature. A third list, l27weak, contained genes of opposite character, i.e., greater importance at the higher, near-optimal temperature. This list contained 917 genes identified by the 1034 dsRNA amplicons that were associated with unusually low cell numbers after knockdown at 27 °C (average B-score ≤ -2) but not after knockdown at 17 °C.

Two additional lists were generated with hits characterized by the opposite, i.e., unusually high rather than low cell numbers at only one of the two tested temperatures. In case of l27strong, cell counts were unusually high at 27 °C; it comprised 738 target genes identified by 816 distinct dsRNA amplicons. The list l17strong contained 966 genes identified by 1078 unique dsRNA amplicons resulting in unusually high number of cells at 17 °C.

Candidate genes important at both temperatures were also identified for comparison with the candidate genes of differential importance at the two analyzed temperatures. Cell counts were unusually low at both temperatures after knockdown of 1014 genes identified by 1162 distinct dsRNA amplicons (average B-score ≤  − 2 at both 17 and 27 °C, see l_both_weak). These 1014 genes of non-temperature-dependent importance were slightly more than the genes primarily important at 17 °C (809) or primarily important at 27 °C (917) (Fig. 2a). In contrast, candidate genes associated with unusually high cell numbers after knockdown at both temperatures (average B-score ≥  + 2 at both 17 and 27 °C, see l_both_strong) were far less numerous than those of differential importance (Fig. 2b).

Validation of selected candidate genes of increased importance at low temperature

A selection of genes of increased importance at low temperature according to the genome-wide screen was validated with additional RNAi experiments. Candidate genes were chosen based on various criteria. Preference was given to strong hits, i.e., those with an extensive difference between cell counts at 17 and 27 °C, as well as low variability between the two same-temperature replicates. However, strong hits were only considered in case of clear evidence for expression in S2R + cells according to RNA-Seq data (Brown et al. 2014). Finally, genes without functional annotations were given low priority.

An initial set of 24 candidate genes, selected early on based on the visual comparison of the overall DNA content in the low magnification images (Fig. 2a), was characterized by low cell numbers specifically at 17 °C. The subsequent quantitative cell count analyses with high-resolution images also identified all these genes as of higher importance at low temperature (i.e., they were present in the l17weak or the l17weak_a hit lists) except for five (Prosα7, Prp6, Phb2, CG5390, Cul1). For experimental validation of the selected 24 candidate genes, an additional RNAi experiment was completed, and knockdown effects at 17 and 27 °C were again compared. For this validation experiment, the same dsRNA amplicon as in the genome-wide screen was used, but the dsRNA was independently prepared. Moreover, some technical aspects of the validation experiments were different (scale of cell culture, image acquisition, and analysis, see “Materials and Methods”). The validation experiment confirmed the results of the screen for 21 of the 24 selected genes at least qualitatively (Fig. 3a). After knockdown of these 21 genes, a lower number of cells were detected at 17 compared to 27 °C. For 14 of these 21 genes, the 27 °C/17 °C cell count ratio was higher than 1.5 (Fig. 3a). The 21 confirmed genes were subject to a second round of validation using distinct dsRNA amplicons for RNAi. In this second experiment, lower cell numbers at 17 compared to 27 °C resulted in case of 15 out of the 21 genes, but a 27 °C/17 °C cell count ratio > 1.5 was observed for only 5 genes (Fig. 3a).

Fig. 3

Validation of screen hits associated with low cell counts after knockdown especially at 17 °C. Candidate genes suggested to be of increased importance at low temperature by the genome-wide RNAi screen were selected based on either overall DNA signal intensities in low-magnification images (a) or cell counts according to high-magnification images (b). For validation experiments, RNAi was induced either with the same dsRNA amplicon as in the screen (dark bars, 1st dsRNA) or with a second distinct dsRNA amplicon (light bars, 2nd dsRNA). After addition of dsRNA, S2R + cells were cultured at either 17 °C for 10 days or 27 °C for 3 days before determination of cell counts. Ratio of cell counts at 27 °C and 17 °C in the tables are marked with light shading when greater than 1.1 and dark shading when greater than 1.5. See Materials and Methods for explanations concerning normalization of cell counts, error bars, and number of replicates

A second set of candidate genes was selected based on the cell count analysis with high-resolution images (Fig. 2b). In total, 20 additional candidates were selected, 11 from the list l17weak and 9 from the list l17weak_a. This second set of candidate genes was also analyzed in an RNAi experiment with an independently produced dsRNA preparation of the same amplicon used before in the screen. For a few candidates, also a distinct additional dsRNA amplicon was used. For 15 of the 20 genes, cell counts were lower at 17 compared to 27° after knockdown with the screen dsRNA amplicon (Fig. 3b). Ten of the 15 genes were characterized by a 27 °C/17 °C cell count ratio > 1.5. In case of the five genes that were further characterized with a second independent dsRNA amplicon, four were observed to have fewer cells at the low temperature, and three had a 27 °C/17 °C cell count ratio > 1.5 (Fig. 3b).

In summary, the validation experiments confirmed the screen results qualitatively for 80% of the selected candidate genes (n = 44) when the same dsRNA amplicon was used for screening and validation. However, the temperature-dependency of cell numbers after knockdown was modest for several candidate genes in the validation experiments. Considering only those with a validated 27 °C/17 °C cell count ratio > 1.5 reduced the confirmation rate to 50%. As the genes were not randomly chosen for validation, these confirmation rates cannot be extrapolated to all genes included in the hits lists. However, in case of strong hits, the screen results appear to be reproducible in 50% of the cases. Additional validation with a second distinct dsRNA amplicon resulted in a further reduction of the confirmation rate. About 42% of the genes (8 out of 19), which had a 27 °C/17 °C cell count ratio > 1.5 in the re-test with the screen dsRNA amplicon, had again a 27 °C/17 °C cell count ratio > 1.5 when analyzed with a second distinct dsRNA amplicon. A failure of confirmation with a second distinct dsRNA amplicon might reflect either off-targets effects of the first dsRNA amplicon or insufficient knockdown efficiency of the second dsRNA amplicon. Overall, we conclude that our genome-wide RNAi screen provides a highly useful basis for selection of candidate genes of temperature-dependent importance for further detailed functional characterization.

Cell cycle profile analysis for genome-wide identification of genes of temperature-dependent importance

The high-content image data acquired in our RNAi screen offers the possibility to use readouts other than cell counts for the identification of genes with a temperature-dependent requirement. Cell counts are not necessarily a most sensitive readout given the design of our screen assay. The number of S2R + cells increased around fivefold at most during the 3-day incubation period at 27 °C used in the RNAi screen. This increase in cell number cannot be completely prevented, even if a given dsRNA effectively targets a gene essential for cell cycle progression, because depletion is not instantaneous. If effective depletion takes about 3 days, a reasonable estimate for many genes, there cannot yet be an effect on cell counts in our screening assay. The same reasoning also applies for depletion at 17 °C. However, compared to cell counts, significant alterations in the cell cycle profile might arise more rapidly after depletion for some genes. Cells might already be largely arrested, in the G1 phase, for example, while still comparable to controls in number at the time of fixation. Therefore, to identify additional genes of increased importance at low temperature, which might have escaped detection based on cell counts, we performed further screen data analyses on cell cycle profiles. Image-based quantitative analysis of the nuclear DNA signal intensities allowed the generation of histograms of DNA signal intensity per nucleus, i.e., the cell cycle profile of the cells in a given microtiter plate well (Fig. 4a). For extraction of quantitative parameters of the cell cycle profile, a cell population model was fitted to the DNA signal intensity histograms. This model assumes that the cell population within a well is composed of three distinct Gaussian sub-populations. In unperturbed negative control cells (Fig. 4a), the first sub-population (P1) corresponds approximately to the cells in the G1 phase of the cell cycle, the second sub-population (P2) to the G2/M cells, and third sub-population (P3) to abnormal hyperploid cells, which are rare in unperturbed cells. In the large majority of the wells analyzed in the genome-wide RNAi screen, the DNA signal intensity histogram and the fitted cell population model was essentially identical to that observed in the negative control wells (Fig. 4 a and b and Online Resource 1, S1 Table). In contrast, inspection of the cell cycle profile of wells treated with dsRNAs depleting well-characterized cell cycle regulators, for example, Cyclin E (CycE), String (Stg)/Cdc25 phosphatase, and Cyclin A (CycA) (Fig. 4b), clearly revealed the expected changes. A massive enrichment of G1 cells was observed after depletion of CycE (Fig. 4b), and depletion of Stg/Cdc25 phosphatase resulted in an enrichment of G2/M cells (Fig. 4b). In case of CycA, which has been demonstrated to result in endoreduplication in the closely related S2 cells after more extended depletion (Rotelli et al. 2019), the onset of endoreplication was detectable (Fig. 4b). Thus, abnormalities in the cell cycle profile can be detected readily by analysis of our high magnification images.

Fig. 4

Identification of genes of temperature-dependent importance based on cell cycle profiles. (a) Integrated DNA signal intensity was determined for each nucleus in a given well, allowing the generation of histograms, i.e., cell cycle profiles (left). The characteristic cell cycle profile from a negative control well is displayed. A model with three distinct Gaussian cell populations was fitted to the observed cell cycle profile (right). The three populations correspond largely to the G1 cells (P1, orange), to the G2/M cells (P2, green), and to the abnormal cells with higher ploidy (P3, red). The cell cycle profile resulting after model fitting is shown in purple and the observed cell cycle profile in blue. (b, c) Comparison of cell cycle profiles observed after knockdown at 17 and 27 °C, respectively, with the indicated dsRNAs. (b) Minimal temperature dependence was observed in negative control wells and after knockdown of bona fide cell cycle regulators. The enrichment of cells with increased ploidy after knockdown of CycA at 17 and 25 °C is indicated (asterisk and bracket). (c) Examples with abnormal cell cycle profiles generated after knockdown at only one of the two analyzed temperatures. (d, e) Validation of the indicated candidate genes important for a normal cell cycle profile at 17 °C but not 27 °C. (d) After knockdown at either 17 or 27 °C, the cell cycle profile was determined by flow cytometry, as shown for Pvr and Dsor1. (e) The temperature dependence of the cell cycle profile was quantified by calculating the ratio indicated along the x-axis. Absence of temperature dependence result in a value of 1 (red line)

The knockdown effects in case of bona fide cell cycle regulators (like CycE, Stg/Cdc25, and CycA) were observed at both 17 and 27 °C (Fig. 4b). Interestingly, however, there were also genes, where knockdown resulted in an altered cell cycle profile predominantly at only one of the two temperatures, at either at 17 °C in case of Pvr and Dsor1 or at 27 °C in case of CG1140 (Fig. 4c).

For identification of genes, for which knockdown resulted in an altered cell cycle profile predominantly at either the low or the high temperature, we generated hit lists using the data extracted after fitting of the sub-population model (see “Materials and Methods”). The primary purpose of these lists (Online Resource 4, S3 Table) was again assistance in the selection of candidate genes to be validated with additional experiments. Eight candidate genes were selected for validation, focusing on those primarily important at the low temperature. Only strong hits with clear evidence for expression in S2R + cells were chosen. For validation, we completed an RNAi experiment analogous to that used for genome-wide screening. The same dsRNA amplicon was used for RNAi as in the screen but with independently generated dsRNA. After cell culture at a larger scale, flow cytometry was used for analyses rather than imaging as in the screen. Beyond negative control experiments (no dsRNA and lacZ dsRNA), several positive controls (dsRNA of CycE, stg, and Cdk1) were included. The cell cycle profiles obtained with the negative and positive control dsRNAs were analogous to those observed in the screen, and importantly they were very similar at 17 and 27 °C (Online Resource 5, S2 Figure). To quantify the effect of temperature on the cell cycle profiles, we determined first the ratio of the size of the G1 and of the G2/M sub-population. These sub-population ratios were determined for the 17 °C and for the 27 °C condition. Thereafter, these rations were compared by quotient formation (Fig. 4e). The resulting 17 °C/27 °C measure was essentially 1 when no dsRNA or lacZ dsRNA was added for RNAi (Fig. 4e), indicating a near identity of the cell cycle profiles at the two analyzed temperatures. In case of CycE dsRNA, the ratio was 1.69 (Fig. 4e), because the G1 arrest was slightly less pronounced at 17 °C. In case of stg dsRNA, the ratio was 0.51, because the G2 arrest was slightly less pronounced at 17 °C (Fig. 4e). In comparison to the controls, most of the candidate genes were characterized by knockdown effects on the cell cycle profile that were far more temperature-dependent (Fig. 4e). As already in the RNAi screen, knockdown of these candidate genes resulted in an enrichment of G2/M cells that was more pronounced at 17 compared to 27 °C. The most pronounced temperature effects were observed for Pvr and Dsor1 (Fig. 4 d and e), while it was minimal in case of Pop2 (Fig. 4e).

The abnormal cell cycle profile resulting after knockdown of Pvr and Dsor1 at 17 °C might reflect a cell cycle arrest during either G2- or M phase. Alternatively, it might arise from a cytokinesis failure starting relatively late during the RNAi treatment period. Indeed, the cell cycle profiles after knockdown of Pvr and Dsor1 at 17 °C were not just similar to the cell cycle profile resulting from knockdown of the M phase inducer stg/cdc25 phosphatase (Edgar and O'Farrell 1990) but also to that obtained after knockdown of pbl (Online Resource 6, S3 Figure), which is essential for cytokinesis (Lehner 1992). For further insight into the cell-cycle arrest resulting from knockdown of Pvr and Dsor1 at 17 °C, we inspected the images obtained in the screen. Images acquired after pbl knockdown revealed many binucleate cells, as expected (Online Resource 6, S3 Figure). In contrast, knockdown of Dsor1 and Pvr at 17 °C did not result in an increased presence of multinucleated cells. Moreover, the cells had an interphase appearance, excluding an M phase arrest. Finally, cell cycle profiles and images also differed from those resulting after CycA knockdown, arguing against endoreduplication. In conclusion, knockdown of Pvr or Dsor1 at 17 °C induces a cell cycle arrest during the G2 phase.

Overall, the analyses based on cell cycle profiles demonstrated that this permits a sensitive identification of genes that are important for normal cell cycle progression primarily at low temperature.

Increased requirement for Ball/VRK protein kinase at low temperature during early embryonic mitoses

Genome-wide RNAi screening with S2R + cells is less laborious than with flies, but the physiological significance of a given hit is not necessarily identical in cultured cells and in the organism. For further evaluation in the organism, we focused on the gene ballchen (ball), which was particularly important for S2R + cell proliferation at low temperature. ball, which codes for the Drosophila VRK protein kinase homolog, was among the screen hits associated with a most substantial difference in cell counts after knockdown at 17 and 27 °C, respectively. Four distinct ball dsRNA amplicons resulted in a far greater reduction in cell numbers at 17 compared to 27 °C (mean B score of the four amplicons at 17 °C =  − 4.48 and at 27 °C =  − 0.25). Moreover, the functions previously attributed to ball were in line with low-temperature effects exposed by time-lapse imaging with early D. melanogaster embryos. On the one hand, this earlier time-lapse imaging (Radermacher 2012) had suggested that the detachment of chromosomes from the nuclear envelope (NE) during condensation at the start of mitosis is sensitive to low temperature. On the other hand, Ball/VRK is known to regulate the release of chromosomes from the NE at optimal temperature (Asencio et al. 2012; Cullen et al. 2005; Gorjánácz et al. 2007; Ivanovska et al. 2005; Molitor and Traktman 2014).

To corroborate the temperature sensitivity of chromosome detachment (Radermacher 2012), we performed additional time-lapse imaging followed by quantitative analyses (Fig. 5). Embryos expressing Lamin-GFP to visualize the nuclear lamina and His2Av-mRFP to visualize chromosomes were analyzed at the presumed optimal temperature (25 °C) and at a low temperature (11 °C). The analysis was focused on progression through nuclear cycle 12, which occurs during the syncytial blastoderm stage. This stage is characterized by the presence of a monolayer of syncytial nuclei just underneath the plasma membrane, a position most optimal for microscopic analysis. Lamin-GFP is an excellent marker for the nuclear periphery not only during interphase but also during the first half of mitosis, because the nuclear lamina depolymerize slowly and incompletely in D. melanogaster. During interphase, His2Av-mRFP signals were indistinguishable at 25 °C and 11 °C. At both temperatures, these signals were distributed rather uniformly throughout the nucleus (Fig. 5). During chromosome condensation at the onset of mitosis 12 (M12), His2Av-mRFP signals increasingly lost their homogenous nuclear distribution. At 25 °C, the condensing chromosomes receded rapidly from the nuclear periphery and contracted towards the interior (Fig. 5). In contrast, at 11 °C, the condensing chromosome remained closely attached to the nuclear periphery at several focal points (Fig. 5). Chromosome detachment from the nuclear periphery occurred eventually also at 11 °C, with a clear delay compared to 25 °C. A metaphase plate was therefore formed at both temperatures (Fig. 5), followed by a normal exit from M phase.

Fig. 5

Low temperature alters the pattern of chromosome condensation in early embryos. Syncytial blastoderm embryos expressing Lamin-GFP and His2Av-mRFP were analyzed by time-lapse imaging at 25 °C and 11 °C, respectively. Single optical sections with nuclei progressing into M12 are displayed. Time (min:sec) is indicated with t = 0 corresponding to the first frame that displayed the final metaphase plate. The bottom row presents larger fields with additional nuclei at the time point when the temperature induced difference in the spatial pattern of the initial chromosome condensation is most clearly apparent. At 25 °C, condensing chromosome detaches immediately from the nuclear periphery and contracts towards the interior. However, at 11 °C, chromosome detachment from the NE occurs with a delay, and thus the interior (arrow) is comparatively devoid of condensing chromosome, and the periphery is characterized by foci of associated condensing chromosomes (arrowhead). Scale bar = 10 μm

The delayed chromosome detachment at the onset of mitosis in syncytial blastoderm embryos observed at 11 °C supported the notion that the requirement for ball function, which promotes the swift release of condensing chromosomes from the nuclear periphery (Asencio et al. 2012; Cullen et al. 2005; Gorjánácz et al. 2007; Ivanovska et al. 2005; Molitor and Traktman 2014), may be particularly high at low temperature. Accordingly, a partial reduction of ball function is predicted to delay the release of condensing chromosomes from the nuclear periphery most strongly at low temperatures. To assess this prediction, we analyzed embryos collected from females with either one or two functional ball+ gene copies. Progression through early embryogenesis, including cycle 12 of the syncytial blastoderm stage, is supported by maternal mRNAs and proteins, as the major activation of zygotic gene transcription proceeds later during cellularization. Therefore, it is the maternal ball genotype that determines how much ball function is present during early embryogenesis. Embryos were collected from mothers heterozygous for the null allele ball2 (Herzig et al. 2014), and these embryos will be designated as ball+_1 in the following. For comparison, ball+_2 embryos were collected from ball2 heterozygous mothers that also carried a copy of g-ball, a transgene with a genomic fragment that promotes full rescue of ball null mutant flies (Herzig et al. 2014). The mothers of both ball+_1 and ball+_2 embryos provided them also Lamin-GFP and His2Av-mRFP for time lapse imaging, which was completed at various temperatures (11, 14, 18, 25, and 29 °C).

Time-lapse imaging of chromosome condensation at M12 onset in ball+_1 and ball+_2 embryos demonstrated that the extent of the delay in the release of condensing chromosomes from the nuclear periphery was highly dependent on both temperature and ball+ function (Fig. 6). Delayed chromosome release was not observed at 25 and 29 °C (Fig. 6a). Further confirmation that high temperatures do not result in a delayed release of condensing chromosomes from the NE was obtained by imaging at 30 °C (data not shown). In contrast, the delayed release was all the more apparent the lower the analysis temperature (Fig. 6a). Importantly, the delay observed at low temperatures appeared to be stronger in ball+_1 compared to ball+_2 embryos (Fig. 6a). Quantitative image analyses corroborated these conclusions concerning the compound effects of low temperature and reduced ball+ gene function on the release of condensing chromosomes from the nuclear periphery. For quantification, the spatial distribution of His2Av-mRFP signals above a threshold intensity was analyzed during entry into M12. These high intensity pixels (HIPs) arise from chromosome condensation. To determine the spatial distribution of the HIPs, nuclei and their equatorial z-sections were detected automatically in the time-lapse image data. The nuclear interior in the equatorial z-section was then subdivided into six concentric onion ring-like segments, and the fraction of HIPs located in these segments was determined (Fig. 6b). Peripheral chromosome condensation is indicated by high HIP fractions in peripheral segments. In contrast, a rapid release of chromosomes from the periphery during condensation is accompanied by high HIP fractions predominantly in central segments. Entry into M 12 at 11 °C was associated with a transient increase of the HIP fraction in the peripheral segments 1 and 2 (Fig. 6b). In contrast, at 25 °C, the increase in HIP fractions occurred preferentially in the internal segments 4–6 (Fig. 6b).

Fig. 6