Generation, analysis, and transformation of macro-chloroplast Potato (Solanum tuberosum) lines for chloroplast biotechnology

Chloroplast biotechnology is a route for novel crop metabolic engineering. The potential bio-confinement of transgenes, the high protein expression and the possibility to organize genes into operons represent considerable advantages that make chloroplasts valuable targets in agricultural biotechnology. In the last 3 decades, chloroplast genomes from a few economically important crops have been successfully transformed. The main bottlenecks that prevent efficient transformation in a greater number of crops include the dearth of proven selectable marker gene-selection combinations and tissue culture methods for efficient regeneration of transplastomic plants. The prospects of increasing organelle size are attractive from several perspectives, including an increase in the surface area of potential targets. As a proof-of-concept, we generated Solanum tuberosum (potato) macro-chloroplast lines overexpressing the tubulin-like GTPase protein gene FtsZ1 from Arabidopsis thaliana. Macro-chloroplast lines exhibited delayed growth at anthesis; however, at the time of harvest there was no significant difference in height between macro-chloroplast and wild-type lines. Macro-chloroplasts were successfully transformed by biolistic DNA-delivery and efficiently regenerated into homoplasmic transplastomic lines. We also demonstrated that macro-chloroplasts accumulate the same amount of heterologous protein than wild-type organelles, confirming efficient usage in plastid engineering. Advantages and limitations of using enlarge compartments in chloroplast biotechnology are discussed.

methods to generate stable transplastomic plants. Only few examples of transient chloroplast transformation in rice (Oryza sativa) have been published 27,28 . Successful chloroplast transformation in key crops, including cereals, will depend on the ability to find more efficient selection (along with selectable marker genes) and tissue culture methods to efficiently regenerate transplastomic plants.
The ability to significantly increase plastid size provides an intriguing avenue with regards to plastid biotechnology. While the small size of plastids has been shown to not be a limiting factor for biolistic transformation, assuming a strong selection system 29 , for alternative methods of plastid transformation (e.g. microinjection), an increase in plastid size would be beneficial if not necessary. For example, the installation of large DNA molecules (synthetic plastid genomes) will likely require alternative transformation strategies.
In higher plants, chloroplast division is regulated by two FtsZ protein homologous (FtsZ1 and 2) 30,31 . FtsZ1 is a tubulin-like GTPase that, in association with FtsZ2 and other division-related components, regulate chloroplast division by participating in the formation of a z-ring at the mid-organelle [32][33][34][35] . Considering that the stoichiometry between FtsZ proteins is important for their correct function in z-ring formation, the most common strategy to increase chloroplast size is through altering FtsZ1 gene expression 30,[32][33][34] . Overexpression of FtsZ1 in Arabidopsis was altered chloroplast division resulting in a reduction of organelle number and drastic increase in size, leading to the term "macro-chloroplast" 36 . Surprisingly, despite this extreme chloroplast phenotype, macro-chloroplast lines were indistinguishable from wild-type lines with regards to plant growth and morphology 30 . However, in a follow-up study in Nicotiana tabacum (tobacco) a significant growth reduction was observed under high-and low-light irradiation in lines with a severe macro-chloroplast phenotype (defined as 1-3 large macro-chloroplasts per cell) 37 . In medium-light irradiation, the growth of all macro-chloroplast lines were indistinguishable from controls 37 . In potato, macro-chloroplast lines produced alterations in both chloroplasts and amyloplasts 38 . Macrochloroplast potato demonstrated morphological changes in starch granule size as well as increased phosphorous content in isolated starch grana compared to wild-type lines 38 . The ability to alter the size and content of nongreen plastids is particularly relevant to plastid biotechnology for food crop improvement, where non-green plastids represent a primary source of nutrition (e.g. chromoplasts of tomato) 39,40 .
While methods to generate macro-chloroplasts have been established for decades, there are few examples of chloroplast engineering using these altered plastids. The first successful transformation of macro-chloroplasts in tobacco, using biolistics, demonstrated a 40% increase in chloroplast transformation efficiency relative to wild-type controls 41 . In addition, researchers demonstrated that homoplasmic macro-chloroplast plants could be regenerated in tissue culture where they accumulated a similar amount of foreign protein compared to traditional chloroplast engineering 42 . Based on this foundational work, we sought to develop macro-chloroplast potato lines specifically for the purposes of evaluating their potential for chloroplast biotechnology. We hypothesized that the use of macro-chloroplast containing leaves would increase the transformation efficiency of potato, providing a superior method for potato chloroplast engineering. Secondarily, we hypothesized that the use of macro-chloroplasts may increase the capacity for protein overproduction without negatively affecting chloroplast function. This would be particularly useful for production of plant bioreactors for high production of foreign proteins.
Specifically, in this work we engineered AtFtsZ1 overexpressing potato lines and selected for macro-chloroplast phenotypes for use in chloroplast biotechnology. The phenotypes of macro-chloroplast lines were characterized, as well as, the efficiency of chloroplast transformation, selection, and regeneration in selected lines.

Results
Generation of AtFtsZ1 transgenic potato lines with enlarged chloroplast morphology. Nuclear genome-transformed potato lines overexpressing AtFtsZ1 under the control of the CaMV 35S promoter were generated by Agrobacterium-mediated transformation. In a first screen, six independent lines were chosen for in-depth analysis from > 200 putative transgenic lines using microscopic analysis of chloroplast morphology in leaf mesophyll-derived protoplasts ( Supplementary Fig. S1). Thereafter, we focused on two independent lines that had large chloroplasts (AtFtsZ1 lines 1 and 2, Fig. S1), which appeared to be vigorous in tissue culture. These two lines were PCR-positive for the transgene (Fig. 1). The presence of the expression cassette integrated into total genomic DNA preparations was verified by amplifying the upstream nptII selection gene along with the downstream AtFtsZ1 transgene (~ 2 kb DNA bands; Fig. 1a). No DNA bands were detected in wild-type samples. In both wild-type and macro-chloroplast lines, DNA bands at similar intensity were obtained for the endogenous loading control actin indicating unaltered amount of genomic DNA in all samples (~ 0.25 kb DNA bands; Fig. 1a). RT-PCRs using primers specific for AtFtsZ1 confirmed transgene overexpression in the two macrochloroplast lines (Fig. 1b). The DNA gel was normalized using RT-PCR reactions for plastome and nuclear internal controls, rbcL and ef1α, respectively (Fig. 1b). qPCR performed on total genomic DNA isolated from leaves, indicated no statistical difference in the copy number of the chloroplast genome between macro line 1 and the wild-type control (p > 0.05; Fig. 1c). On the contrary, a slight reduction was observed in line 2 (p < 0.05; Fig. 1c).
Top leaves of the canopy from 3-week-old in vitro grown macro-chloroplast lines 1, 2 and wild-type controls were observed using confocal microscopy. In leaf mesophyll cells of transgenic lines 1 and 2, significantly larger and elongated chloroplasts were observed relative to wild-type lines ( Fig. 1d-f, respectively). The same lines were used for ultrastructural characterization of chloroplast morphology by transmission electron microscopy (TEM). Micrographs of ultrathin sections from mesophyll leaf cells confirmed a larger and elongated chloroplast morphology in AtFtsZ1 transgenic potato lines (TEM; Fig. 1d,e) compared to wild-type control plants (TEM; Fig. 1f). Internal chloroplast structures, including thylakoid membranes, appeared to be morphologically normal without any apparent structural alterations ( Supplementary Fig. S2).
The enlarged chloroplast phenotypes in AtFtsZ1 transgenic lines were confirmed by measurement of organelle size from single plane confocal images from isolated protoplasts (Fig. 2a-c). Compared to chloroplast size in wild-type potato leaves (6.7 ± 2.1 µm), the organelle size in macro-chloroplast lines 1 and 2 were substantially compared to wild-type controls was investigated in controlled growth studies with time-point analysis conducted at anthesis and at harvest (Fig. 3). Macro-chloroplast lines demonstrated a 1 week delay in reaching anthesis when compared to wild-type controls ( Fig. 3a,b, respectively). Further, at anthesis, the two macro-chloroplast lines were ~ 22% shorter ( Fig. 3d) than wild-type controls. Despite this decreased growth, neither line had a decrease in total dry weight when compared to the wild-type control ( Fig. 3e and Supplementary Fig. S3a). Despite similar reduction in height between the macro-chloroplast lines, line 1 had significantly more total dry and fresh weight per plant compared to line 2 alone ( Fig. 3e and Supplementary Fig. S3a). For both macrochloroplast lines, chlorophyll content and CO 2 assimilation were significantly lower than the wild-type lines (Fig. 3g,h, respectively). Reduction in these performance metrics was expected to decrease tuber production due to a lack of available carbon. To evaluate this hypothesis, growth parameters and tuber yield were analyzed at harvest (24-weeks). Images of 12-weeks-old plants (prior to senescence) and tubers harvested at 24-weeks are shown ( Fig. 3c). At the end of the life cycle, prior to harvest, the total above-ground biomass is severely affected by extended leaf senescence, thus only plant height and both dry and fresh weight per unit of leaf area were collected. No statistical difference was observed for these growth parameters when comparing macro-chloroplast and wild-type lines (Fig. 3i, j and Supplementary Fig. S4a). Tuber production between individuals was highly variable, leading to no significant difference in tuber number between macro-chloroplast and wild-type lines (Fig. 3k); however, there was a significant reduction in tuber biomass from macro-chloroplast lines ( Fig. 3l and Supplementary Fig. S4b). A reduction of ~ 34% and 45% in total dry and a reduction of ~ 31% and 41% in total fresh weight were observed in macro-chloroplast line 1 and 2, respectively ( Fig. 3l and Supplementary Fig. S4b).
Biolistic transformation of chloroplast genomes in macro-chloroplast lines. The possibility to transform macro-chloroplasts was investigated by using chloroplast transformation vectors delivered by biolistics. Two vectors, pIR and pSSC specific for recombination into the IR (trnI/trnA) or SSC (ndhG/ndhI) regions of plastome were used ( Fig. 4a,b, respectively). Considering that the chloroplast morphology and growth parameters of the two AtFtsZ1 transgenic lines were similar (Figs. 1, 2, 3), only the macro line 1 was used for biolistic   Fig. S5 and S6). In wild-type potato, the transformation efficiency for both constructs was higher than macro-chloroplast lines (Table 1). Primers specific for the trnI/trnA or ndhG/ndhI integration site along with Southern blots were used to check operon integration in the second round of pIR and pSSC transplastomic lines, respectively ( Fig. 4c-f, Supplementary Figs. S5 and S6). PCRs specific for the internal plastome control rbcL and the two transgenes (GFP and aadA) were conducted (Fig. 4c,d and Supplementary Fig. S7). The presence of 2.6 kb DNA-bands at the predicted molecular weight indicated correct operon integration in both plastome regions (trnI/trnA or ndhG/ndhI) in all wild-type and macro-chloroplast transplastomic lines (Fig. 4c,d). In Southern blot analysis the presence of bands at the predicted molecular weight for the trnI/trnA and ndhG/ndhI integration sites with the cassette integrated (3.8 kb and 4.5 kb, respectively Fig. 4e,f) confirmed correct integration. In Southern blots for both pIR and pSSC lines, the absence of wild-type bands (1.6 kb and 2.3 kb, respectively, Fig. 4e,f) indicates 100% homoplasmy. Compared to normal and macro-chloroplast transplastomic lines, there was no difference in the level of homoplasmy for either integration site (trnI/trnA or ndhG/ndhI, Fig. 4e,f).
GFP accumulation in macro-chloroplast lines. GFP fluorescence was detected using the confocal microscope in chloroplasts in wild-type and macro-chloroplast lines transformed with pIR or pSSC constructs. These fluorescent profiles confirmed GFP production and correct subcellular localization in all transplastomic lines (Fig. 5a-d). The GFP amount produced in leaf tissue was also investigated by fluorometry (Fig. 5e). Quantification of GFP indicated that both normal and macro-chloroplast transplastomic lines accumulated the same amount of protein (ng of GFP) per mg of leaf fresh weight (Fig. 5e). No statistically significant difference was observed in GFP accumulation in both genotypes when compared to transplastomic plants transformed with the same construct, pIR or pSSC. Considering the IR region is present in two copies in the chloroplast genome, both wild-type and macro-chloroplast transplastomic lines transformed with pIR accumulated more protein then pSSC plants (~ 128 and 186% increase, in normal and macro, respectively). In order to investigate if these levels of GFP expression were enough for efficient standoff detection, transplastomic pIR and pSSC lines of both    (Fig. 5f,g). Fluorescent images of normal and macro-chloroplast lines transformed with pIR or pSSC indicated bright green fluorescence detected in engineered plant canopies at a 3-m standoff (Fig. 5f,g, respectively). The growth of wild-type and macro-chloroplast transplastomic lines was investigated in real-time for 9 weeks on soil (Supplementary Video S1). As expected, the transplastomic pSSC lines had faster growth compared to the macro-transplastomic pSSC lines. After 9 weeks the transplastomic pSSC lines reached a similar size to wild-type (untransformed) controls, whereas macro-chloroplast lines had reduced height. On the contrary, both transplastomic and macro-transplastomic pIR lines had significantly reduced growth due to the high output of heterologous protein production (Fig. 5e). The results are expressed as mean ± standard deviation (SD) of 2 biological replicate and 2 technical replicate per biological replicate per each independent line. A number of 15 pIR and 8 pSSC lines were used per each genotype, respectively (N normal; M macro). Means were compared using ANOVA and when significant, mean separations were analyzed using Tukey HSD (p < 0.05). Statistical significance is indicated by numerical letter (a,b). Images acquired using the FILP (Fluorescence-Inducing Laser Projector) stand-off detection system of transplastomic lines along with wild-type controls (WT) are shown: 3-week-old pIR and pSSC lines of both genotypes (N and M) on potting are shown in f and g, respectively. GFP signal, bright field (BF) and merge images are indicated in (f) and (g). The images were acquired at 150 ms exposure time. Scale bar: 50 mm (f,g).

Discussion
The ultrastructural characterization of leaf tissue from potato AtFtsZ1 lines using TEM showed the presence of large and elongated chloroplasts throughout all analyzed cells (Fig. 1, 2 and Supplementary Figs. S1, S2). This is in accordance with previous results demonstrating that the overexpression of FtsZ1 affects the normal formation of the z-ring, resulting in inhibition of division and excessive chloroplast growth 36,38,42 . Analyzing TEM micrographs, we did not observe any evidence for ultrastructural differences of thylakoid membranes in macro-chloroplasts compared to wild-type organelles ( Fig. 1 and Supplementary Fig. S2). The results from this work suggest that macro-chloroplast lines have a delay in growth and significant reduction in tuber biomass compared to wild-type controls ( Fig. 3 and Supplementary Fig. S3). Despite slower growth, the macro-chloroplast plants had no apparent changes in leaf and stem morphology compared to wild-type controls (Fig. 3). The delay in growth was likely due to reduced photosynthetic performance as supported by reduced amount of chlorophyll and CO 2 assimilation in macro-chloroplast lines (Fig. 3g,h). Altered growth characteristics may also be the result of altered leaf optical properties in macro-chloroplast lines. It has been demonstrated in tobacco AtFtsZ overexpressing lines, the particular chloroplast phenotype and cell rearrangement can lead to either decreased light absorbance (at low-light irradiation) or photodamage of the photosynthetic machinery (at high-light irradiation), resulting in reduced plant growth in both cases 44 .
The prospect of using agronomically-relevant plants with large organelles is an attractive addition to the plastid biotechnology toolkit. Macro-chloroplasts from AtFtsZ1 transgenic lines can be transformed using biolistics and transplastomic lines can be regenerated using standard tissue culture. We anticipated that increasing the size of the target organelle may provide an advantage in transformation efficiency in potato as it has been previously demonstrated in tobacco 41 . Considering the transformation efficiencies of potato chloroplasts reported in the literature [45][46][47] , our macro-chloroplast lines can be efficiently transformed, yielding approximately 1 transplastomic plant per each plate bombarded (0.9 and 1.1 for pIR and pSSC construct, respectively, Table 1). However, the transformation efficiency of transgenic lines was not greater than the efficiency obtained in our laboratory for wild-type potato leaves. For both constructs tested (pIR and pSSC) about 4 positive plants per each plate bombarded (3.5 and 3.9, respectively, Table 1) were obtained. This observation suggests that there is not a clear advantage of having larger organelles using biolistics. This low transformation efficiency could be due to the particular spatial distribution of macro-chloroplasts in leaf mesophyll cells. Compared to wild-type, macro-chloroplasts seem not to be regularly distributed throughout the cytoplasm and are instead pushed to the plasma membrane ( Fig. 1d-f). This particular distribution may reduce the probability of gold particles hitting the target organelle.
Nevertheless, this result does not exclude the potential for macro-chloroplasts to be desirable targets for other gene-delivery methods in which the transformable surface is an important limiting factor. For almost 3 decades, the gene gun has been the most efficient and reproducible method used for chloroplast transformation 48,49 . This is an invasive physical method that could induce the shearing of large vectors and, thus, is not particularly suited for transformation of large, chromosome-sized constructs. The effective transformation of large DNA molecules, such as an entire synthetic plastome, may necessitate alternative methods of transformation. Many methods including micro-injection 50,51 , polyethylene glycol (PEG) transformation 52,53 and electroporation 54,55 have been used to transform chloroplasts. Different kinds of commercial carbon-nanotubes have also been used to deliver DNA into plants cells 56,57 , and these delivery techniques could also represent an alternative to deliver vectors more efficiently into macro-chloroplasts. Thus, macro-chloroplasts could provide considerable advantages in increasing transformation efficiency using these methods.
Compared to conventional transplastomic lines, macro-chloroplast lines do not differ in the level of homoplasmy for both trnI/trnA and ndhG/ndhI integration sites (Fig. 4). Two rounds of selection in tissue culture were required to get pIR and pSSC homoplasmic lines (trnI/trnA and ndhG/ndhI) in both macro-and wild-type chloroplast-engineered plants. While a difference in plastome copy number was previously observed in macrochloroplast and wild-type tobacco 58 , our results suggest that potato AtFtsZ1 overexpressing lines have similar plastome copy number (Fig. 1c). Further, similar heterologous protein (GFP) production was observed in wildtype and macro-transplastomic lines and varied by construct/integration site, as expected (Fig. 5e). Based on growth analysis of pSSC transplastomic and macro-transplastomic lines, both reached full maturity, while pIR transplastomic and macro-transplastomic lines demonstrated significantly stunted growth due to the increased production of heterologous protein. These results suggest that the amount of heterologous protein production is a much more significant consideration than chloroplast size when conducting chloroplast biotechnology. Moreover, as observed for wild-type chloroplasts, the level of GFP accumulation obtained in macro-chloroplast lines was enough for efficient standoff imaging of whole plant canopy using the FILP system (Fig. 5f,g). This is of particular interest in a context of modern agriculture, where phenomics has the prospect to bring together high-dimensional phenotypic and molecular data together with their interaction with environmental stimuli 43,59 .
For determination of growth characteristics, apical shoots (same length and age) from both transgenic and wild-type plants were in vitro propagated. After 2 weeks, small plantlets with roots were transferred to soil, Pro-Mix BK25 (Griffin Greenhouse Supplies, Inc. Tewksbury, MA, USA), and grown to anthesis (9-and 10-weeks-old for wild-type and AtFtsZ1 lines, respectively) or full maturity (24-weeks-old for both genotypes). Pots of 3.8 or

Scientific Reports
| (2020) 10:21144 | https://doi.org/10.1038/s41598-020-78237-x www.nature.com/scientificreports/ 11.4 l volume were used to grow plants at anthesis or full maturity, respectively. Plant height (cm), fresh and dry weight (g) of leaves and total biomass were measured. The total number, along with the dry and fresh weights (g) of tubers were also measured. For determination of dry weight, plant tissue was dried for 1 week at 50 °C. Both leaf and tuber areas were obtained by image analysis using ImageJ 1.41o software (National Institute of Health, Bethesda, MD, USA). CO 2 assimilation (A) values per unit of leaf area (µmol m −2 s −1 ) were obtained using a LI-6800 portable photosynthesis system (LI-COR Biosciences, Lincoln, NE, USA) at atmospheric CO 2 concentrations (400 µmol mol air −1 ), constant irradiance (1000 µmol photons m −2 s −1 ), ambient temperature of 23 °C, a vapour pressure deficit (VPD leaf) of 0.8-1.2 kPa and a flow rate of 200 µmol s −1 ). The leaf chlorophyll content index (CCI) was obtained using a portable CCM-200 plus chlorophyll content meter (OPTI-SCIENCES Inc., Hudson, NH, USA).
Results were expressed as mean ± standard deviation (sd) of six and four biological replicates for each genotype (AtFtsZ1 independent lines (Macro 1 and 2) along with wild-type control) for the growth experiments at anthesis and full maturity, respectively. ANOVAs with post-hoc Tukey statistical analysis (IBM SPSS software) (p < 0.05) were performed to determine if differences were statistically significant among wild-type and AtFtsZ1 lines. Plants grown in vitro and in pots were subjected to congruent environmental conditions, 16/8 h of light and dark respectively, and the temperature was kept constant at 24 °C.
Generation of transgenic lines. Potato AtFtsZ1 overexpressing lines (macro-chloroplast lines) were generated by Agrobacterium-mediated transformation. The vector pAP202 36 was introduced into Agrobacterium tumefaciens LBA4404 by the freeze thaw method 62 . For each transformation, 20 internodes of ~ 1 cm-length from 1-month-old in vitro potato plants were used. Internodes were then transformed via Agrobacterium and transgenic plants were regenerated in selective media, as previously described 63 . Putative transgenic lines were screened by fluorescent microscopy for clearly-defined enlarged chloroplasts. Thereafter, selected lines were genetically characterized for construct integration and transgene expression.
The PDS-1000/He biolistics device (Bio-Rad, Hercules, CA, USA) was subsequently used to transform chloroplasts of both macro-chloroplast and wild-type potato plants 64 . Macro-chloroplast line 1 was used for chloroplast transformation. Per each transformation ~ 6 cm 2 of leaf tissue from 1-month-old in vitro potato plants were used. Per each shoot 0.3 mg of gold-particles (0.6 µm in diameter) binding 1 µg of plasmid were used following the manufacturer protocol (Seashell Technology, La Jolla, CA, USA). Transplastomic plants of both genotypes (wild-type and macro-chloroplast) were then regenerated from transformed leaf material incubated in selective media as described previously 60 . The second round of transplastomic plants were obtained by applying the same protocol of tissue culture/selection/regeneration 60 . All lines were genetically characterized for the presence of the selection cassette integrated and GFP expression.
Total DNA extraction and PCR analysis. Total genomic-DNA preparations from leaves were obtained using ~ 50 mg of tissue and the CTAB-based procedure of extraction 64 . PCR (25 cycles) was performed in 25 µl reaction-volume by using 5 ng of total genomic DNA and the DreamTaq Green PCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA). For molecular characterization of AtFtsZ1 overexpressing lines, the pairs of primers 2Fw/2Rv and 3Fw/3Rv were used to check for nptII/AtFtsZ1 cassette integration and the internal control actin (GeneID-102593904), respectively. For genotyping of transplastomic lines, the two pairs of primers 4Fw/4Rv and 5Fw/5Rv were used to verify cassette integration in the trnI/trnA and ndhG/ndhI sites of plastome, respectively. The two pair of primers 6Fw/6Rv and 7Fw/7Rv were used to amplify full-length GFP (NCBI ID: AEX93343.1) and aadA (NCBI ID: AAR14532.1) genes, respectively. The primers 8Fw/8Rv were used to amplify an internal portion of rbcL (NCBI ID: 4099985).
Total RNA extraction and reverse transcriptase PCR. Total RNA extraction was performed using Tri-Reagent (Molecular Research Center, Inc, Cincinnati, OH, USA) according to manufacturer's protocol. Per each extraction ~ 50 mg healthy leaf tissue was used. Total RNA preparations were cleaned and subjected to DNase treatment using the RNA Clean & Concentrator Kit (Zymogen, Irvine, CA, USA) according to manufacturer's instruction. The cDNA synthesis was performed using the Super Script III Reverse Transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) following the manufacturer's instruction. cDNA was quantified using a NanoDrop (Thermo Fisher Scientific, Waltham, MA, USA), and several serial cDNA dilutions were tested by PCR (as described before) for the presence of the transgene AtFtsZ, whereas the two internal controls rbcL (plastome) and ef1α (nuclear) were used for normalization. The pair of primers 9Fw/9Rv, 10Fw/10Rv and 11Fw/11Rv were used to amplify and internal fragment (~ 100 bp) of AtFtsZ (Tair ID: AT5G55280), rbcL (NCBI ID: 4,099,985) and ef1α (NCBI ID: NM_001288491.1), respectively.

Scientific Reports
| (2020) 10:21144 | https://doi.org/10.1038/s41598-020-78237-x www.nature.com/scientificreports/ Southern blot analysis. DNA probes for detection of pIR and pSSC constructs integrated into potato plastome (GenBank: NC_008096.2) were designed on IR (104,457-104,978 bp) and SSC (120,269-120,790 bp) regions, respectively. The PCR DIG Probe Synthesis Kit (Roche, Indianapolis, IN, USA) was used to synthesize digoxigenin(DIG)-sUTP-labelled IR and SSC DNA-probes using the pair of primers 12Fw/12Rv and 13Fw/13Rv, respectively. Total genomic DNA from leaf tissue was extracted using CTAB, as described above. After quantification, 1 µg of DNA for each pIR and pSSC sample was digested using KasI/HindIII and FspI/ScaI restriction enzymes, respectively. The DNA fragments were separated on 0.9% agarose gel, and after that, the gel was depurinated, denatured and transferred on a nylon membrane as described previously 13 . The membrane was then incubated with the DIG-labelled probe and detected using the anti-digoxigenin-AP Fab fragments detection kit (Roche Indianapolis, IN, USA) accordingly to the manufacturer's protocol.  Table S1). The QuantStudio 6 Flex Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA) was used to perform real-time PCR, whereas the QuantStudio Real-Time PCR Software v1.1 (Thermo Fisher Scientific, Waltham, MA, USA) was used to aquire amplification data. The data were expressed as Log 2 (2 −ΔCT ) of rbcL vs actin. Results were expressed as mean ± standard deviation (sd) of 3 biological replicates and 8 technical replicates per biological replicate per each genotype. An ANOVA post-hoc Tukey statistical analysis (p < 0.05) was performed to determine statistically significant difference among means (IBM SPSS software).

GFP quantification.
The amount of GFP accumulated in leaf tissue of transplastomic lines was quantified by using the Fluorometric GFP Quantification Kit (Cell Biolabs, Inc., San Diego, CA, USA) according to the manufacturer's protocol 66 . The results were expressed as mean ± standard deviation (ng GFP/mg fresh weight) of two indipendent experiments, resulting in a total of 2 biological replicates and 2 technical replicates per biological replicate for each independent transplastomic line. We analyzed 15 pIR and 8 pSSC lines for each genotype, normal and macro-chloroplast.
Confocal microscopy and chloroplast size determination. Leaf tissue from 3-week-old in vitro plants was imaged using an Olympus Fv1200 confocal microscope (Olympus, Center Valley, PA, USA). GFP was excited at 488 nm and detected at an emission wavelength of 509 nm. Chlorophyll autofluorescence was visualized using an excitation wavelength of 543 nm and an emission wavelength of 667 nm. Confocal images were acquired using the manufacturer's Olympus FV10-ASW Viewer software Ver.4.2a (Olympus, Center Valley, PA). Confocal images were processed using the online software ImageJ 1.41o (National Institute of Health, Bethesda, MD, USA). The same ImageJ software was also used to estimate chloroplast size in both AtFtsZ1 overexpressing lines and wild-type controls. The results were expressed as mean ± standard deviation (sd) and the statistical analysis was performed using SPSS statistics 25 software (IBM).
Transmission electron microscope (TEM). Leaf tissue from 3-week-old in vitro plants were chemically fixed in glutaraldehyde/ paraformaldehyde along with an osmium tetroxide solution and embedded in EMBed-821 resin (Electron Microscopy Sciences, Hatfield, PA, USA) as described previously 67 . Ultrathin sections (60-100 nm) of embedded tissue were cut using a diamond knife and a Leica EM UC7 ultramicrotome (Leica, Buffalo Grove, IL, USA). Ultrathin sections were post strained using aqueous solutions of uranyl acetate and lead citrate as described previously 67 . Images were obtained using a JEOL 1400 transmission electron microscope operating at 80 kV (JEOL, Peabody, MA, USA) equipped with a Gatan OneView camera (Gatan, Pleasanton, CA, USA). TEM micrographs were processed using the software ImageJ 1.41o (National Institute of Health, Bethesda, MD, USA).

Fluorescence imaging by the fluorescence-inducing laser projector (FILP). Plant fluorescence
patterns were characterized using the Fluorescence-Inducing Laser Projector (FILP), a custom-built instrument for imaging fluorescence in plants. FILP imaging was performed on 3-week-old transplastomic lines in pots 43 . GFP fluorescence in plants was acquired at 150 ms exposure time using 465 nm excitation and 525 nm emission notch (50 nm) filter. Standoff detection was performed at 3 m from the laser source. FILP images were processed using ImageJ 1.41o software (National Institute of Health, Bethesda, MD, USA).