Establishing Tg (cyclin B2:Luc/GFP) MITO-Luc/GFP transgenic zebrafish lines
To develop a zebrafish model to visualize proliferation events in the whole animal, the promoter fragment used in MITO-Luc reporter mice22 was cloned upstream the sequences of firefly luciferase (Luc) and GFP reporter genes within the zebrafish Tol2 transposable element pT2KXIG plasmid (Fig. 1A). We used the mouse promoter since the three NF-Y subunits are highly conserved between mouse and zebrafish34(Supplementary Table S1). Moreover, zebrafish cyclin B2 promoter contains two of the three CCAAT boxes present on the fragment of the murine promoter22(Supplementary Fig. S1), thus indicating that murine cyclin B2 promoter region may be sensitive to NF-Y activity also in zebrafish. First, we tested weather the resulting pT2KXIGΔin-cyclin B2-Luc/GFP plasmid was actually a read out of proliferation. To this aim we transiently transfected with pT2KXIGΔin-cyclin B2-Luc/GFP plasmid, a mouse myoblast cell line that allows to easily observed mitotic and post-mitotic states35 as confirmed by a immunofluorescence microscopy confocal analysis after 3 h BrdU pulse (Supplementary Figure S2). In vitro luciferase activity assay showed 14.5-fold increased transcriptional activity in mitotic cells compared to post-mitotic ones (Supplementary Figure S3) thus proving the specificity of this reporter system for proliferating cells. To investigate the activity of pT2KXIGΔin-cyclin B2-Luc/GFP construct in zebrafish it was transiently injected into one to four-cell stage wild type embryos and, as control, the pT2KXIGΔin-Luc/GFP empty construct was transfected, too. In vitro luciferase activity assay performed in lysates showed 80-fold increased transcriptional activity in pT2KXIGΔin-cyclin B2-Luc/GFP compared to pT2KXIGΔin-Luc/GFP injected embryos (Fig. 1B). These experiments demonstrate a murine cyclin B2 promoter-dependent luciferase expression in injected zebrafish embryos.
To generate stable lines, the pT2KXIGΔin-cyclin B2-Luc/GFP construct and Transposase mRNA were co-injected into one to four-cell stage wild type embryos. Transgenic founders were collected by using GFP fluorescence microscopy analysis. Two independent transgenic zebrafish lines were identified and named MITO-Luc/GFP1 and MITO-Luc/GFP2. Quantitative PCR analysis indicated a different number of the transgene copies in the two lines (Fig. 1C). Consistent with these results, MITO-Luc/GFP1 expresses more in vitro luciferase activity than MITO-Luc/GFP2 embryos (Fig. 1D) and in vivo confocal microscopy shows a higher fluorescence level in MITO-Luc/GFP1 vs MITO-Luc/GFP2 embryos (Supplementary Fig. S4); finally, Southern Blotting assay confirmed multiple transgene insertions in MITO-Luc/GFP1 (Supplementary Fig. S5). In contrast, inverse PCR36 in MITO-Luc/GFP2 (Supplementary Fig. S6) indicated a single transgene insertion located in the first intron of the AMP-activated, gamma 2 non-catalytic subunit protein kinase gene (prkag2a) on chromosome 24, position (28.173.305–28.302.316)(ZFIN Gbrowse). These results are consistent with the quantitative PCR data (Fig. 1C). PCR on cDNAs obtained from transgenic lines show that both luciferase and GFP transcripts are present on a single transcript (Supplementary Fig. S7). On the contrary western blot on protein extracts from transgenic embryos show that the proteins are two (Supplementary Fig. S8). Sequence of the construct shows that the two open reading frames are separated by two CC (Supplementary Figure S9) suggesting that a leaky scanning translation occurs on the transcript, leading to luciferase and GFP proteins expression from the same mRNA37.
Fluorescence and bioluminescence correlate with proliferation in MITO-Luc/GFP zebrafish strain
To evaluate the reliability of MITO-Luc/GFP zebrafish lines as a new tool to visualize cell proliferation through GFP fluorescent signal, we focused on the MITO-Luc/GFP1 which expresses the higher number of inserted transgenes. FACS analysis performed on cells derived from about 200 transgenic embryos at 40 hpf, subjected to a 3 h BrdU pulse, showed a high predominance of BrdU accumulation (Fig. 2A) in GFP positive gated population rather than in GFP negative gated population (Fig. 2B), indicating that in this transgenic strain GFP is mostly expressed in proliferating cells. As expected, not all GFP positive cells have incorporated BrdU. This is due to the half-life of the GFP protein which, being 26 h38, also traces cells that have stopped to proliferate in the previous 26 h.
Next, GFP expression was observed at 6, 24 and 33 h post fertilization (hpf) embryos by fluorescent microscopy (Fig. 2C). At early stages (Fig. 2C I,II) the GFP expression is almost ubiquitous throughout the embryo; in contrast, it appears to be tissue specific in 33 hpf embryo where it is detected in the head (Fig. 2C III, red arrow), proximal trunk (Fig. 2C III, white arrows) and in developing fin (Fig. 2C III, yellow arrow). A background signal coming from the yolk sac is due to auto-fluorescence as demonstrated in wild-type fishes (Supplementary Fig. S10). Interestingly, this picture recapitulates published data on the recruitment of BrdU in 36 hpf embryos exposed to pulses of this DNA analog39.
Next, using dechorionated and low-melting agarose embedded live 4,5 hpf embryos, sequential stages of embryonic development were captured by confocal microscopy (until 50% epiboly stage)40. GFP-positive cells were identified along the entire acquisition (Supplementary Fig. S11) and different cell layers were distinguished (Supplementary movie S1 online). Morphogenetic cell movements of extension that characterize the early development of zebrafish embryos from 30 to 50% epiboly were also detected (Supplementary Fig. S11). Of note, from fertilization to 50% epiboly stage, cells divide continuously thus indicating that GFP positive are proliferating cells.
Similarly, 19 hpf embryos were imaged until the 26 somites stage. In Supplementary movie S2 online, somites development and eye enlargement are visible and origin of the primary organs and constriction of yolk extension are apparent. At this stage several cells are dividing rapidly: the brain and lens develop, neurons and muscle precursors establish and the tail elongates40, confirming that GFP positive are proliferating cells.
Additional experiments were performed to establish whether physiologic proliferation events could be visualized in whole animals. Embryos, juveniles and adults were collected, anesthetized, bathed in D-luciferin and in vivo imaged (Fig. 3A–C). One embryo per well was imaged at 24 hpf stage. BLI demonstrated a signal from the entire embryo, as expected at this stage of development, when all cells are highly proliferating (Fig. 3A). Of note, this result confirms GFP expression analysis at the same stage. In juvenile and adult animals light is much more specifically localized in the abdomen (Fig. 3B,C). Luciferase enzymatic activity was also measured in homogenates from embryo, juvenile and adult MITO-Luc/GFP1 zebrafish; AB wild-type strain was used as negative control (Supplementary Fig. S12). The in vitro results closely resemble those seen in vivo (Supplementary Fig. S13), being luciferase activity higher in embryo than in juvenile and adult homogenates. Unexpectedly, luciferase activity was more pronounced in juvenile than in adult fish. This result can be explained considering proliferation/total tissue ratio. In fact, being juveniles smaller than adults, in in vitro experiments, proliferating cells are less diluted in the homogenates from the entire animal. This does not happen in vivo, where only the signal coming from proliferating tissue is detected.
To elucidate which abdominal organs are the source of light seen in vivo by BLI, photon emission was measured ex vivo, in explanted organs from both male and female adult fishes (Fig. 3D). High luciferase activity was detected in gonads, where a high number of proliferating cells are present41,42. Although to a lesser extent, also other organs produce light: brain, kidney and intestine (these organs are known to include a low number of proliferating cells)43,44,45,46. As expected, heart is negative and comparable to background, since, in adult animals, under baseline conditions, myocardial cell proliferation does not occur.
Markers of proliferating cells co-localize with luciferase and GFP in MITO-Luc/GFP zebrafish tissues
To verify the suitability of MITO-Luc/GFP1 zebrafish model as tracer of proliferation events the cellular distribution of luciferase and GFP were analysed by immunofluorescence. As expected, both reporters are highly expressed in almost all nuclei in the distal trunk section of 24 hpf embryos, being in this stage all cells highly proliferating (Supplementary Fig. S14). Likewise, in the trunk region of 24 hpf whole-mount transgenic embryos (Fig. 4A), several GFP positive cells are co-expressing the proliferation marker phosphorylated histone 3 isoform (pH3) a well-known marker of mitosis47(Fig. 4A II,III,VII).On the contrary, lack of overlapping signals between GFP and the post-mitotic neurons marker HuC/D is clearly observable (Fig. 4A III, IV, VIII). Overall, these data indicate that, in MITO-Luc/GFP zebrafish embryos, GFP is mostly expressed in proliferating cells while post-mitotic cells, like those expressing HuC/D marker, are all GFP negative.
Furthermore a colocalization of luciferase and GFP was observed also in adult epidermal fin tissue (Fig. 4B I,II,V) thus demonstrating that the two reporters are expressed in the same cells. Interestingly, several GFP positive cells also express pH347(Fig. 4B III,IV,VI). In line with these results, the majority of GFP positive cells in intestine (Fig. 4C I,II,III) and in ovary (Fig. 4C IV,V,VI) also express proliferation cellular nuclear antigen (PCNA), another well-documented marker for cell proliferation in zebrafish48. Overall, these data indicate that GFP positive are proliferating cells while post-mitotic cells are GFP negative.
Although the majority of GFP positive cells are also positive for proliferation markers, not all of them are. One possible explanation is due to the longer half-life of GFP protein (26 hr)38 compared with that of luciferase49 and PCNA50. Regarding pH3 this is a specific mitotic marker47, thus limiting the percentage of colocalization with our tracer GFP expressed along all cell cycle phases.
Next, by Immunohistochemistry (IHC), sequential paraffin section from adult MITO-Luc/GFP1 tissues were stained with antibodies against luciferase, PCNA or pH3. Of note, in all analyzed tissues, cells are found that express proliferation markers, either PCNA or pH3, and also immunoreactivity for luciferase. Luciferase and PCNA are coexpressed in the majority of the cells of the intestinal villi (Fig. 4D I-II) and testis spermatogonia (Fig. 4D III-IV). Luciferase and pH3 are coexpressed in several cells of the nephron proximal tubule in kidney (Fig. 4D V-VI) as well as in gill filaments and lamellae (Fig. 4D VII-VIII). As expected, in this last case co-expression is not 100% as pH3 is a mitotic phase marker of the proliferating cells, as discussed above (Fig. 4D IX).
Taken together, these results demonstrate that both luciferase and GFP reporter genes are expressed in proliferating cells in MITO-Luc/GFP1 zebrafish, and suggest that this animal model is a powerful tool to visualize cell proliferation in live animals.
Caudal fin regeneration in MITO-Luc/GFP zebrafish line
These experiments were aimed at establishing whether cell proliferation could be visualized after inducing tissue damage in MITO-Luc/GFP zebrafish. For these experiments we took advantage of zebrafish ability to regenerate amputated fins, a process associated with cell proliferation51. Cell proliferation and migration contribute to the early regeneration of zebrafish fins; epithelial and mesenchymal cells proximal to the level of the amputation, few hours later the injury, are strongly labeled with BrdU52. The expression of GFP was imaged during regeneration of the caudal fin in 3 days post fertilization (dpf) MITO-Luc/GFP1 embryos after fin clip. Embryos were subjected to the cut of the distal segment of the caudal fin, embedded in agarose and imaged for 14 h (Supplementary movie S3 online). Imaging demonstrated growth of the blastoma (in Fig. 5A selected frame pictures from the movie). A rapid migration of GFP positive cells to the wound site is observed, thus indicating that these cells are proliferating to replace the wounded epidermis with new tissue. This result was confirmed by immunofluorescence analysis, where the recruitment of GFP positive cells on blastoma, 1 day after cut, is clearly detected (Fig. 5B). In agreement, 24 h after a caudal fin fragment dissection from MITO-Luc/GFP2 2 dpf live embryos, many GFP positive cells are visible at the tip of the caudal fin, adjacent to the region of amputation (Fig. 5C). This experiment confirmed that, in both transgenic zebrafish line, GFP is expressed only in cells involved in proliferative events, such as caudal fin regeneration.
Next, we monitored the progress of caudal fin regeneration by BLI in adult animals after fin clip procedure. Three adult zebrafish males and three females were anesthetized and bathed in D-luciferin and ex vivo imaging of the caudal fin was performed; the animals were sacrificed 1 and 7 days after fin clip. Luciferase activity of the fin stump doubled 1 day after fin clip and was back to its baseline value one week after the procedure (Fig. 5D). We investigated this process also in vivo by BLI by serial imaging of the same animal following fin clip. After the amputation, fishes were placed on a water-soaked sponge and subjected to longitudinal in vivo imaging sessions; a representative experiment is shown in Fig. 5E. Images were obtained before fin clip (Fig. 5E, precut). The first day after the cut, the caudal fin stump emits intense light indicating that cell proliferation is occurring in that area (Fig. 5E 1 day, red arrow), whereas the emitted light decreases on the subsequent days. A week post amputation light emission detected in the regenerated fin has completely ceased (Fig. 5E 7 days). These observations are in agreement with published data on caudal fin regeneration showing that the wound healing process starts approximately 12 to 48 h post-amputation51 even if amputated fin grows for longer than a week to reach the original length, in the last part of the process few cells are proliferating and might not be seen, due to a matter of sensitivity of the model.
In parallel with the signal coming from caudal fin, we observed a systemic luciferase activation in the abdomen and head starting the first day after amputation (Fig. 5E 1 day, white arrow) and persisting at least for two additional days (Fig. 5E 2 days, 3 days). Light emission at all sites ceased and was completely undetectable one week after fin clip (Fig. 5E 7 days).
To further verify which organ/s emit/s light, ex vivo experiments were carried out to measure luciferase activity of dissected organs from three males and three females sacrificed 1, 3, and 7 days after fin clip. Luciferase activity was measured in dissected organs previously bathed in D-luciferin. Although to a different extent, high luciferase activity was detected in intestine, brain, heart and kidney dissected 1 day after fin clip (Fig. 5F). The gonads exhibited a slight increase in luciferase activity after fin clip as well, but these changes were not statistically significant (Supplementary Fig. S15).
Taken together, our data demonstrate that after fin clip, GFP and luciferase expression are induced in MITO-Luc/GFP1 zebrafish strain and strongly indicate that in this model system it is possible to dynamically visualize induction of proliferation events in specific organs of interest.
Inhibition of luciferase activity in zebrafish embryos upon anti-proliferative treatments
We investigated whether inhibition of cell proliferation was associated with inhibition of luciferase expression in MITO-Luc/GFP1 zebrafish model. To this aim, a pool of 24 hpf embryos was treated with a sub-lethal dose of a well-known anti-proliferative drug, 5Fluoro Uracil (5FU)53. We acquired images of the embryos just before treatment (Fig. 6A, pre). 5FU was dissolved in water containing dechorionated 24 hpf embryos and replaced with fresh water after 6hrs (Fig. 6A, 0 h). Treated embryos were then analyzed in vivo through imaging sessions at sequential timing. After 6, 18 and 24 h treatment we observed a strong inhibition of luciferase expression compared to untreated embryos (Fig. 6A, 6 h, 18 h, 24 h). The signal of 5FU-treated embryos progressively increased after removal of the drug and, after 42 h it was comparable to control (Fig. 6A, 42 h). Luciferase in vitro enzymatic assay, performed on embryos lysates, confirmed the in vivo results (Fig. 6B).
In additional experiments, 24 hpf embryos were irradiated using sub-lethal X-rays at different doses54. Luciferase activity of untreated, 1.8 Gy and 2.7 Gy treated embryos was measured in vitro, 6 h after irradiation. As shown in supplementary figure S16, X-ray radiation determined a reproducible decrease of luciferase activity in irradiated groups, compared with not-irradiated controls.
To further support the hypothesis that MITO-Luc/GFP zebrafish may be useful in large-scale drug screening experiments, the effect of Etoposide and Nocodazole, two cell cycle inhibitors, was investigated in 24 hpf embryos treated with sub lethal doses of these drugs55,56. We found that both cell cycle inhibitors markedly decreased luciferase activity compared to untreated control embryos (Fig. 6C I,II). As expected, both the experiments depicted in Fig. 6B and 6C show a time-dependent decrease of luciferase expression in untreated embryos, a result due to the progressive decrease of proliferative events during the time course of these studies.
Taken together these results demonstrate that anti-proliferative treatments decrease luciferase activity in MITO-Luc/GFP1 zebrafish, suggesting that this animal model represents a useful tool to screen the efficacy of candidate anti-proliferative drugs.