Real-time detection of somatic hybrid cells during electrofusion of carrot protoplasts with stably labelled mitochondria

Somatic hybridisation in the carrot, as in other plant species, enables the development of novel plants with unique characteristics. This process can be induced by the application of electric current to isolated protoplasts, but such electrofusion requires an effective hybrid cell identification method. This paper describes the non-toxic fluorescent protein (FP) tagging of protoplasts which allows discrimination of fusion components and identification of hybrids in real-time during electrofusion. One of four FPs: cyan (eCFP), green (sGFP), yellow (eYFP) or the mCherry variant of red FP (RFP), with a fused mitochondrial targeting sequence, was introduced to carrot cell lines of three varieties using Agrobacterium-mediated transformation. After selection, a set of carrot callus lines with either GFP, YFP or RFP-labelled mitochondria that showed stable fluorescence served as protoplast sources. Various combinations of direct current (DC) parameters on protoplast integrity and their ability to form hybrid cells were assessed during electrofusion. The protoplast response and hybrid cell formation depended on DC voltage and pulse time, and varied among protoplast sources. Heterofusants (GFP + RFP or YFP + RFP) were identified by detection of a dual-colour fluorescence. This approach enabled, for the first time, a comprehensive assessment of the carrot protoplast response to the applied electric field conditions as well as identification of the DC parameters suitable for hybrid formation, and an estimation of the electrofusion success rate by performing real-time observations of protoplast fluorescence.

Breeding new F1 hybrid carrot cultivars relies on the use of male sterile lines. Male sterility in the carrot is determined by cytoplasm, thus a mitochondria transfer to a new genomic context is required for the development of new CMS lines. This is commonly done by repetitive backcrossing which is a difficult and long-lasting process, in particular in the carrot, an allogamous and biennial crop exhibiting strong inbreeding depression 57 . Protoplast hybridisation as an alternative approach may considerably shorten the time of cytoplasm transfer, however, its utilization requires the initial establishment of an efficient and reliable system confirming the successful transfer of mitochondria. Hence, the aim of this work was to develop a model allowing the confirmation of mitochondria transfer between protoplasts and the testing conditions favouring electrofusion. For this purpose, we developed a set of carrot lines whose cells had fluorescently labelled mitochondria by different FPs, and used them as sources of protoplasts for somatic hybridisation. We demonstrate that real-time observation of protoplast fluorescence during electrofusion enables the assessment of the effect of various DC parameters on protoplast stability and hybrid formation, and the optimisation of these parameters for somatic hybridisation. Moreover, for the first time, observations of the dual-colour fluorescence emitted by hybrid cells have directly confirmed a successful mitochondria transfer and enabled the estimation of the electrofusion efficiency in the carrot.

Results and discussion
Carrot cell transformation. Embryogenic cell suspensions of three varieties were used as convenient targets for genetic transformation ensuring effective exposure of plant cells to A. tumefaciens 58 . The microscopic observations of the ' Amsterdamska' suspension in the medium containing ammonium glufosinate conducted 3 days after transformation revealed the presence of elongated single cells and cell divisions. After 7 days, the suspension became denser and 1-3 mm micro clusters of cells were distinguished as was expected in the event of successful transformation 59 . Such changes in the ʻKoralʼ and DH suspensions were delayed, and cell aggregates were visible 12 days after transformation. Microscopic observations revealed that cell fractions resistant to ammonium glufosinate and capable of further divisions and development were present in all suspensions. The fluorescence observed 8-10 days after transformation was detected in all 12 cell suspensions (3 varieties × 4 FP genes). The non-transformed suspension cultured in the BI medium without a selection agent continued to grow but the cells did not emit fluorescence. Neither divisions nor fluorescence was found in the negative control, i.e. the non-transformed cell suspension treated with ammonium glufosinate.
The obtained suspension cultures consisted of non-fluorescing and fluorescing cells of various fluorescence intensities. Clear detection of fluorescence requires a high level of FP gene expression, which can be ensured, for example, by the 35S promoter, as in this work. It also requires the synthesis of FP peptide, which undergoes correct folding and maturation 60,61 . Disturbances in these processes may result in the lack of fluorescence or fluorescence quenching. A transient gene expression may also lead to unstable fluorescence. The models derived for carrot cells expressing GFP showed that weakening fluorescence was a function of time, which additionally depended on the carrot cultivar used as the cell source, and the fraction of fluorescing cells stabilised about a week after transformation 48 . For these reasons, the obtained suspensions in this work were composed of fluorescing cells, putatively transgenic cells not showing fluorescence and untransformed cells that survived a short period of selection. However, the identification of fluorescing cells indicated that the cell suspensions originating from the cells of the carrot storage root can be expedient for genetic transformation that had been questioned earlier 62,63 .

Development of callus with FP-labelled cells. The occurrence of non-fluorescing cells in transgenic
cell suspensions implied the need for a two-step selection using a solidified medium supplemented with the selection agent to develop homogenous lines of fluorescing cells. The formation of small aggregates on a solidified medium was observed first for the ʻAmsterdamskaʼ cells, and the process was noted just 5 days after the suspension transfer onto a filter paper disc placed on the medium surface. In the next 3 weeks, aggregates covered the filter paper, but then only some of them continued their growth and callus clumps became distinguishable. A similar development, although at a slower rate, was obtained for the cultures originating from the ʻKoralʼ and the DH line, which was congruent with the slower growth of these two varieties in the cell suspension culture, and confirmed the essential role of the source material used for transformation. Callus did not develop on the selection medium when non-transformed cells were cultured. The use of filter paper discs had been proposed earlier for an easy transfer of cells and their small aggregates to fresh nutrient media 59 , and was then adopted for selection purposes 64 . This procedure allowed repeated, undisturbed transfer of all cells and their exposure to the selection agent until the developing callus clumps could be separated and placed directly on the surface of fresh medium; hence, it increased the chance for the development of transgenic tissue. Four 6-week-old putative transgenic calli, growing well, from each treatment were sampled for PCR analysis. The amplification of the bar and FP gene fragments resulted in products of the expected lengths listed in Table 1 and confirmed the presence of each FP gene. Thus, molecular analyses provided evidence for successful gene transfer and selection of transgenic events in the callus.
All PCR positive callus lines emitted fluorescence; however, the proportion of fluorescing cells varied considerably among the materials and did not usually exceed 50% as estimated during a brief microscopic observation. At high magnification (400×), numerous, small, and scattered fluorescing spots were clearly identified within fluorescing cells (Fig. 1). The set of plasmid vectors used for transformation in this work contained a signal molecule targeting FPs to mitochondria and comes from the collection created by Nelson et al. 56 , who used it for fluorescent labelling of membrane-bounded organelles in Arabidopsis cells. Therefore, the fluorescing spots observed in cytoplasm could be identified as labelled mitochondria. The callus with introduced YFP emitted bright fluorescence, allowing for easy discrimination between fluorescing and non-fluorescing cells. Slightly

Scientific Reports
| (2020) 10:18811 | https://doi.org/10.1038/s41598-020-75983-w www.nature.com/scientificreports/ lower fluorescence intensities were observed for GFP and RFP calli. Clear identification of single mitochondria tagged with CFP was not possible, as most CFP cells displayed faint fluorescence. The lower intensity of CFP fluorescence may be partially explained by the lower molar extinction coefficient and the lowest quantum yield of the ECFP variant in comparison to other FPs used. As a consequence, the ECFP brightness expressed in relation to the brightness of the reference EGFP is 39%, while for sGFP, EYFP and mCherry, the relative brightness values are 160%, 151% and 47%, respectively 65 . Despite the lower brightness, CFP has been useful in co-localisation studies using multicolour organelle labelling 66 . The callus lines, including CFP, with the highest proportion of fluorescing cells and the highest fluorescence intensity were chosen for further culture and the second step of selection (Table 2). A brief inspection of fluorescence was conducted before every subculture to favour the transfer of callus fragments exhibiting the most   www.nature.com/scientificreports/ intense fluorescence. In consequence, after the next 6 weeks of culture, the proportion of fluorescing cells in the developing calli increased (Table 3). However, in all CFP calli, CFP fluorescence remained faint; the scoring of fluorescing cells was ambiguous and clear discrimination between fluorescing and non-fluorescing cells was highly subjective. Hence, the CFP calli were excluded from further experiments despite their growth on the selection medium and positive molecular verification.

Protoplast isolation from FP-labelled callus lines.
Depending on the callus line, the isolation efficiency using a modified washing protocol was 1.2-2.6 × 10 6 protoplasts per -gram of callus, which was in the range expected for a well working isolation procedure applied to tissue in a good physiological condition 6,67,68 . The obtained protoplasts were spherical, indicating complete enzymatic digestion of the cell wall and their diameter ranged from 22 to 33 μm depending on the line ( Table 3). The cell size variation was noticeable among the callus lines originating from the same carrot variety but, on average, DH protoplasts (23.3 µm) were smaller (P < 0.001) than the ' Amsterdamska' and 'Koral' protoplasts (28.2 µm and 33.0 µm, respectively). The DH callus was obtained from the roots of the DH carrot line, which was bred through the reproduction of a doubled haploid plant achieved after doubling the haploid chromosome set 69 . In contrast to other varieties, the DH line is thus completely homozygous and characterised by a smaller leaf and storage root size that is common for carrot inbred lines, which usually exhibit strong inbreeding depression 57,70 . The smaller cell size of DH line may result from inbreeding depression due to complete homozygosity. The percentage of protoplasts emitting fluorescence was high (Fig. 2); in most 12-week-old callus lines, it was above 80% up to 95% (Table 3). These values were, on average, 2-3 times higher than the estimated percentages in a 6-week-old callus. Although direct comparison of fluorescence of isolated protoplasts in suspension and in callus cells should be done with caution, the obtained percentages indicate successful selection of fluorescing callus fragments done during consecutive subcultures and on the development of calli composed mainly of Table 3. Mean protoplast diameter and percentage of fluorescing protoplasts isolated from 12-week-old callus. N number of protoplasts. 1 Means followed by the same letter do not differ at P = 0.05 according to the Tukey's test for protoplast diameter and Chi-square test for percentage of fluorescing protoplasts.  Optimisation of electrofusion parameters. Real-time observations of the protoplasts isolated from the YFP-labelled 'Koral' callus allowed identification of distinguished electrofusion stages described earlier by Navrátilová 71 and Hu et al. 28 . At first, the protoplasts were randomly scattered between two electrodes. Their dispersion occurred due to mutual repulsion of the negatively charged protoplast membranes until the alternating electric field was switched on. Then the adhesion of protoplasts forced by dielectrophoresis caused protoplast alignment in pearl chain structures. The application of DC caused reversible cell membrane perforation and, for some protoplasts, coalescence with the adhering protoplasts. Finally, after 3-5 s of membrane re-stabilisation, round hybrid protoplasts were formed. The hybrids of two cells had diameters larger than the donor protoplasts by about 25%, which was expected when the volume of a new hybrid protoplast was doubled.
Membrane integrity and formation of hybrids highly depends on parameters of the applied electric field 32,72,73 hence, the effect of fifteen combinations of DC voltages and pulse times on hybrid formation was assessed. In general, the cells remained intact, and no fusion was observed at low voltage and with short pulses (Fig. 3). Gradually increasing voltage above 2 kV/cm or pulse time above 40 μs led to fusion, but in most cases, the fusion was incomplete. At high values of these parameters, hybrids were formed but they were unstable and eventually disintegrated. Setting the parameters to the highest voltage (3.5 kV/cm) and longest pulse (100 μs) caused irreversible protoplast damage. Complete fusion and formation of stable hybrids was observed when applying intermediate voltage values (2.5 and 3.0 kV/cm) and pulse times (50 or 60 μs). These observations are consistent with the thesis that shorter pulses of higher voltage are preferable in successful protoplast electrofusion 74 . As the result of screening, these 4 out of 15 combinations of current parameters were considered to favour hybrid formation.
Real-time identification of heterofusants. When somatic hybridisation aims to combine protoplasts that have different genetic backgrounds, the identification of heterofusants and distinguishing them from homofusants is crucial. In this work, the protoplasts were labelled with three FPs, GFP and YFP, whose spectra partially overlap 75 , and RFP which emits fluorescence in longer wavelengths 76 . Hence, two component combinations, RFP + GFP and RFP + YFP, were chosen to detect heterofusants. Real-time observations during electrofusion allowed the differentiation of protoplasts labelled with the FPs emitting either red or green/yellow fluorescence. After the electrofusion process was completed, it was possible to identify protoplasts emitting fluorescence in two spectra (Fig. 4). Observation of dual-colour fluorescence indicated that such protoplasts were composed of two fused components, each labelled with different FPs (RFP + GFP or RFP + YFP); hence, they were identified as heterofusants. An inspection in transmitted light was carried out to verify whether there were any other pro- Fusion efficiency. Eight pairs of callus lines were used for protoplast electrofusion and four combinations of voltage and pulse time were set, based on the above-described optimisation experiments carried out using 'Koral' protoplasts. Additionally, a lower DC of 2 kV/cm was applied for 50 and 60 µs to extend the range of electrofusion parameters including those that were considered less harmful. The percentage of obtained heterofusants ranged from 0.9 to 13.9%, but at the highest voltage (3.0 kV/cm) or a longer pulse (60 µs) the fusion was often unsuccessful ( Table 4). The efficiency highly depended on the pair of fusion components (P < 0.001) and their interaction with DC voltage and pulse (P < 0.001); this 3-way interaction explained one third of total variation. Two-way interactions between fusion components and DC voltage or DC pulse explained another 26% of variation (both P < 0.001). On average, the efficiency doubled by increasing DC voltage from 2.0 to 2.5 kV/ cm, but only when the pulse lasted 50 µs (Fig. 5). Further voltage increase to 3.0 kV/cm favoured heterofusant formation for two line combinations and was adverse for the remaining five combinations. At the longer pulse (60 µs), changes to voltage did not substantially affect the mean efficiencies.   www.nature.com/scientificreports/ The combination of callus lines used as fusion components affected the heterofusion efficiency (P < 0.001) and highly contributed (30%) to total variation. The highest efficiency (8.7% ± 1.01 s.e.) observed for the DH-RFP + DH-GFP pair was eight times greater than for the A-RFP + A-GFP pair (1.1% ± 0.51 s.e.). On average, the combination of protoplasts isolated from two DH callus lines labelled with different FPs more often resulted in successful heterofusant formation (P = 0.005; n = 24), with a mean efficiency of 5.5% ± 0.9 s.e., than when the protoplasts were isolated from the two ' Amsterdamska' callus lines (2.3% ± 0.8 s.e.). The DH protoplasts were smaller than the ' Amsterdamska' protoplasts, however, the protoplast size did not explain the observed differences in the effectiveness of heterofusant formation. When protoplasts of similar size, 22.7 µm (DH-GFP) and 22.3 µm (DH-YFP), were used for fusion in combination with DH-RFP, the mean efficiencies (8.7% and 2.2%, respectively) differed significantly (P < 0.001; n = 12). Thus, the results do not support the conclusions from other reports 31 that the use of protoplasts of similar size favours hybridisation and are congruent with those questioning such relationships and explaining the differences in fusion response with different metabolic characteristics of fused cells 72 . The DC voltage and DC pulse main effects were also statistically significant (P = 0.001 and P = 0.013, respectively) but their contribution to the total variation was only 3%.
Considering both the efficiency of heterofusant formation and the positive response of various materials to electrofusion, the DC parameters set to 2.5 kV/cm and 50 µs were the most effective. At these parameters, heterofusants were obtained in all experiments for all pairs of components, and the efficiency ranged from 1.3 to 8.2%, with a mean of 4.3% ± 0.65 s.e. To the best of our knowledge, the exact electrofusion parameters are given in only one report describing the development of cybrid carrot plants after fusion of suspension-derived protoplasts using two DC pulses of 50 µs at 1.0 kV/cm, however, the results concerning the efficiency were not presented 21 . The efficiency obtained in this work cannot be directly compared, as there is no such data available for the carrot. In other species, also with the help of different set of morphological markers, the estimation of heterofusants frequency is difficult and time consuming, and hence such results have been rarely presented. In Brasicaceae, the frequency of heterofusant formation after PEG-and electric field-induced fusion ranged from 0.5 to 10% 71 , while in the citrus species, electrofusion led to 1-5% of heterofusants 27 . Higher values (11-22%) were noted during electrofusion between pea and grass pea protoplasts 77 . The results presented in this work indicate that electrofusion of callus-derived carrot protoplast may lead to heterofusant formation with satisfactory efficiency in comparison to other species.
The efficiencies estimated here did not account for the fact that electrofusion might additionally lead to homofusion. Homofusants could not be identified by observation of dual-colour fluorescence, as they emitted fluorescence in only one spectrum range. An indirect confirmation of homofusant formation was the enlargement of some protoplasts after electrofusion and their fluorescence in one spectrum range only. Quantification of the frequency of this process would be, however, highly biased due to protoplast variation in size observed before fusion. Assuming the probability of homo-and heterofusion was similar, as the protoplasts of both components were mixed in a 1:1 ratio, the mean efficiency would be equal to a doubled frequency of heterofusant formation i.e., above 8%.
In general, electrofusion is considered advantageous over the PEG-mediated fusion due to its low cell toxicity, but the latter approach is preferably chosen, independent of the plant species 28 . Electrofusion requires the use of special equipment, such as pulse generator, and the optimisation of alternating current and DC parameters, which must be adjusted depending on the species, cell type and cell size. Papers describing the effect of all these factors are rare and they concern the electrofusion of protoplasts derived mainly from mesophyll and cell suspensions 27,72 . The reported ranges of applied DC parameters are wide. Mostly, DC voltages ranged from 1.0 to 3.5 kV/cm, pulses lasted for 40-100 µs, and DC was applied once or repeated 2-3 times, when using pairs of the same or different cell types as fusion components. These components included suspension/suspension-leaf/ leaf-or suspension/callus-derived protoplasts [78][79][80] . In the present work, the selected DC parameters ensuring the highest efficiency of carrot hybrid formation fit in these ranges, although a shorter pulse time was more advantageous.

Conclusions
The presented results have shown that carrot protoplasts can be stably labelled by FPs targeting mitochondria and, for the first time, that such fluorescent markers can be useful for immediate discrimination of parental components during the whole process of protoplast fusion, stimulated by the application of electric current. They have also demonstrated that the hybrids can be identified due to the emission of dual-colour fluorescence. Despite avoiding the need for additional protoplast staining, the protoplast response to applied conditions varied greatly, indicating that the physiological condition of fusion components remains one of the most critical factors in somatic hybridisation. Moreover, the optimal conditions favouring fusion for one pair of components could not be as good as for another pair, and there was no direct relationship between hybridisation efficiency and protoplast size or protoplast origin. Nonetheless, some combinations of DC parameters ensured successful fusion of any pair of components and the obtained hybrid cells underwent divisions; hence, they should be selected for further somatic hybridisation in the carrot using other protoplast sources.

Material and methods
Biological material. Well established, 4-6-month-old, embryogenic cell suspensions derived from storage roots of three carrot (Daucus carota L. ssp. sativus Hoffm.) varieties were used: a doubled haploid DH1 line (DH) 69   www.nature.com/scientificreports/ surface of sterile filter paper disks placed on a 0.27% Phytagel (Sigma) solidified BI medium in Petri dishes supplemented with 200 mg/L cefotaxime, 100 mg/L timentin and 10 mg/L ammonium glufosinate as a selection agent. The filter paper discs were transferred to fresh media every 2 weeks, and then the developing, individual callus clumps were directly placed on BI selection medium. The callus was cultured at 26 °C in the dark and was subcultured every 2 weeks to a fresh medium with the addition of 10 mg/L ammonium glufosinate. The callus clumps obtained from the non-transformed carrot cell suspensions were considered as negative controls. The remaining part of the suspension not transferred to Petri dishes was further cultured on the gyratory shaker after supplementation with 5 mg/L ammonium glufosinate.

PCR.
Genomic DNA was isolated from 6-week-old callus using the CTAB method according to the protocol by Rogers and Bendich 86 with modifications as described by Klimek-Chodacka et al. 87  Monitoring of the post-fusion cell development. After DH-RFP and DH-GFP protoplasts electrofusion at the selected optimal DC conditions, the protoplasts were embedded in thin alginate layers and cultured in the protoplast culture medium according to a previously described protocol by Grzebelus et al. 68 . The protoplast viability before and just after electrofusion was verified by fluorescein diacetate (FDA) staining 68 . Additionally, the formation of cell aggregates was observed in 7-, 10-and 20-day-old cultures, and the percentage of cell aggregates was determined in relation to the total number of observed cells/aggregates.
Microscopy. Aliquots of cell suspensions or protoplasts were directly placed on a microscopic slide. Pieces of callus were submerged in distilled water first, and then aliquots of loosened cells were transferred to a microscopic slide. The preparations were examined using the inverted Zeiss AxioObserver A1 microscope with fluorescence mode (Carl Zeiss) equipped with a set of High Efficient (HE) filters dedicated to observation of fluorescence. For the FP variants, the following filters were used: CFP-HE47 (excitation wavelength λ Ex = 436/25, emission wavelength λ Em = 480/40), GFP-HE38 (λ Ex = 470/40, λ Em = 525/5), YFP-HE46 (λ Ex = 500/25, λ Em = 535/30) and RFP-HE63 (λ Ex = 572/25, λ Em = 629/62). The cells and protoplasts were observed in bright field mode using 200× magnification. During electroporation, the protoplasts were observed in transmitted light (bright field mode) and in reflected light (fluorescence mode) after excitation with a mercury lamp. Each specimen was observed using HE filters changed according to the FP present in the cells. The images were captured using the AxioCam MRc 5 camera (Carl Zeiss) attached to the microscope. To observe a mixture of protoplasts possessing mitochondria labelled with different FPs and to detect hybrid protoplasts emitting dual-colour fluorescence after fusion, the recorded images taken using different HE filters were merged using ImageJ2 × 2. www.nature.com/scientificreports/ line) and expressed as the percentage of all visible cells/protoplasts in transmitting light. The Chi-square test with Yates correction was used to identify significant differences between proportions of fluorescing protoplasts. After electrofusion, the efficiency of heterofusant formation was estimated by calculating the percentages of protoplasts emitting dual-colour fluorescence and dividing these percentages by the proportions of fluorescing protoplasts determined for fusion components before electrofusion. This correction was applied to take into account for fluorescence heterogeneity in source materials. A three-factor ANOVA (pair of fusion components, DC voltage and DC pulse time) was performed after Bliss transformation of percentage data. Significance between means was verified using the Tukey's test. Two fusion experiments were performed for each factor combination using protoplasts from independent isolations (96 fusions in total) and at least 100 cells per combination were observed. The assessment of protoplast viability and colony formation in cell culture after electrofusion at the optimal DC conditions was carried out on 200-350 cells per each time point in four repetitions and data were subjected to one-way ANOVA followed by the Tukey's test.