Arabidopsis (Arabidopsis thaliana, At) Columbia accession (Col-0) was used for all experiments. A transgenic line expressing CHUP1-NT-GFP was selected among T3 lines with normal chloroplast positioning, based on green fluorescent protein (GFP) fluorescence around the chloroplast outer envelope (Oikawa et al., 2008). The line was used to examine the delivery of the KH-AtOEP34–DNA complex (cpPD complex) to the chloroplast outer membrane and stroma. For plasmid DNA delivery and integration of exogenous DNA into chloroplasts, plants were grown for 2months under a 8-h light/16-h dark short-day photoperiod on soil to gain a large leaf.

KH-AtOEP34 [protein sequence: (KH)₉-MFAFQYLLVM] was previously synthesized and purified (Yoshizumi et al., 2018). The purity and the molecular weight of the peptide were confirmed using high-performance liquid chromatography with an InertSustain C18 column (GL Science, Tokyo, Japan) and through matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (Yoshizumi et al., 2018), respectively. We have previously constructed the plasmid PpsbA-aadA-sGFP-TpsbA (Yoshizumi et al., 2018). In this study, we used the plasmid for DNA delivery as part of the complex with the peptide KH-AtOEP34. DNA fragments for the psbA promoter (PpsbA) and psbA terminator (TpsbA) were amplified with the primer sets (5'-GAAGATCTGCAAGAAAATAACCTCTCCTTC-3' and 5'-CTCCTCGCCCTTGCTCACCATTCTCTCTAAAATTGCAGTCATGGTAAAATCTTGG-3') for psbA promoter and (5'-GAAAAATTCTATAGAAACTTCTCTCAATTAGGATCCTGGCCTAGTCTATAGG-3' and 5'-GAAGATCTGCAAGAAAATAACCTCTCCTTC-3') for psbA terminator. The DNA fragment PpsbA:sGFP:TpsbA was generated by combining each PCR fragment and cloned into the pUC19 vector (Yoshizumi et al., 2018). We also used the plasmid Prrn-aadA-sfGFP-Trps for integrating exogenous DNA into Arabidopsis plastid genome for GFP expression analysis. The plasmid was constructed following the methods (Lutz et al., 2007). Prrn gene is a sequence for promoter of plastidic ribosomal RNA, aadA-sfGFP gene is a sequence coding aminoglycoside adenyltransferase fused to sfGFP, and Trps is a terminator sequence of plastidic ribosomal RNA 16S. In the plasmid, 5' and 3' flanking regions of the Prrn-aadA-sfGFP-Trps region contains DNA sequence for transfer RNA L (trnl) and transfer RNA A (trnA) for recombination with chloroplast genome, respectively. The plasmids used in this study are listed in Supplementary Table S1.

To construct KH-AtOEP34–DNA complexes in a 0.5N/P ratio (with N being the number of amine groups from the peptide and P being the number of phosphate groups from the plasmid DNA), plasmid DNA was purified with the Qiagen DNA extraction kit. After an incubation of 1min at 23°C and thorough mixing, the KH-AtOEP34 peptide was added to the DNA solution and incubated for 1h at 23°C (Midorikawa et al., 2019; Miyamoto et al., 2019).

To visualize KH-AtOEP34–DNA complex, the plasmid DNA was labeled with cyanine fluorescent dye (Cy3) using a Nucleic Acid Labeling Kit (Label IT tracker Reagent, Mirus). After purification of the plasmid DNA using ethanol precipitation, 5μg of plasmid DNA conjugated to Cy3 in 100μl was mixed with KH-AtOEP34 at an N/P ratio of 0.5 (Chuah et al., 2015; Chuah and Numata, 2018).

The peptide–DNA complex were characterized with a zeta potentiometer (Zetasizer Nano-ZS; Malvern Instruments, Ltd., Worcestershire, United Kingdom) following published methods (Chuah et al., 2015). Each mixture of cpPD complex was prepared to a final volume of 800μl using ultrapure water (Milli-Q) at an N/P ratio of 0.5. The mixture was immediately placed into a cuvette to measure the zeta potential and the size with a zeta potentiometer, and the averaged data were obtained using Zetasizer software version 6.20 (Malvern Instruments, Ltd.).

Around 100μl of peptide–DNA complex solution was infiltrated with a syringe into the abaxial side of a leaf of a 2-month-old Arabidopsis plant grown in short-day conditions (Chuah et al., 2015). The leaves were then cut into four pieces and observed by CLSM at different times after infiltration.

Fluorescence signals from the peptide–plasmid DNA complex labeled with Cy3 (cpPD-Cy3 complex) or GFP localized at the chloroplast outer membrane were detected using the CLSM (Zeiss LSM880) with a 63x oil immersion objective (Plan-Apochromat 63x/1.4 Oil DIC M27) and a zoom (x 1.5–2.5) in Airy scan mode (Carl Zeiss, Jena, Germany) with an Argon and DPSS laser for excitation of GFP and Cy3 and the following excitation/emission wavelengths: 488/495–550 nm for GFP and 561/495-620 nm for Cy3. Time-lapse images were taken every 0.5–5s for 120–600s and stacked as a movie file using Imaris (Carl Zeiss, Jena, Germany) or Image J (Fiji; Schindelin et al., 2012) after taking single images from the surface to bottom of the mesophyll cell. The plasma membrane of leaf sections was stained with 20μM FM4-64 dye (Thermo Fisher, United States) for 10min after infiltration of the cpPD-Cy3 complex for 1–12h. The leaf sections were then rinsed in distilled water once before observation by confocal microscopy using excitation/emission wavelengths: 561/565–590nm for Cy3, 561/620–650nm for FM4-64 dye, and 488/660–700nm for chlorophyll autofluorescence. Chloroplast DNA was stained with SYBR Green I (Dragan et al., 2012) for 30min after infiltration of the cpPD-Cy3 complex for 6h. SYBR Green I fluorescence was observed by confocal microscopy using excitation/emission wavelengths: 488/490–540nm for SYBR Green I, 488/660–720nm for chlorophyll autofluorescence, and 561/565–590nm for Cy3.

Field emission-scanning electron microscopy (FE-SEM; Gemini 3000, Zeiss, Germany) was performed following a previous method (Sogawa et al., 2020). A sample of the peptide–DNA complex (cpPD complex) was prepared as described above. The cpPD complexes were enveloped in 2% (w/v) agarose and sectioned to 1–2mm for FE-SEM observation. Leaf sections treated with the cpPD complex were embedded in epoxy resin and sectioned to 100μm for FE-SEM observation.

The method followed to count the number of peptide–plasmid DNA complexes labeled with Cy3 (PD-Cy3 complex) inside and outside leaf mesophyll cells is described in Supplementary Figure 1. The ratio of cpPD-Cy3 inside and outside cells was calculated. The frequency of the Phase I-IV was calculated as each number of the Phase I-IV chloroplasts divided by the total number of the chloroplasts in the cell, which has the chloroplasts stained with SYBR Green I with the cpPD-Cy3 complexes. The frequency of the cpPD-Cy3 complexes bound to the cp-nucleoids was calculated as divided by the total number of the cpPD-Cy3 complexes on the chloroplast.

To understand how the peptide-pDNA complex is internalized into a cell and delivered to chloroplast nucleoids, we tracked the peptide-pDNA complex using imaging technology with the previously designed peptide, KH-AtOEP34, and the plasmid PpsbA-aadA-sGFP-TpsbA (Yoshizumi et al., 2018). The chimeric peptide KH-AtOEP34 consists of a DNA-binding cationic peptide and a signal peptide for targeting to the chloroplast outer membrane, in complex with the plasmid to assess gene delivery to the chloroplast (Figure 1A; Yoshizumi et al., 2018). The negative charge of the plasmid DNA allowed its complexation with the cationic peptide (Figure 1A). The successful delivery of these peptide–DNA (cpPD) complex entails the penetration of the cell wall and plasma membrane, followed by delivery to the chloroplast and DNA integration into the chloroplast genome, as shown in Figure 1B.

We first determined the hydrodynamic size and surface charge (Z-potential) of the cpPD complexes, following a method that differed slightly from the one reported previously (Yoshizumi et al., 2018) to stabilize the size of the cpPD complexes (Midorikawa et al., 2019; Miyamoto et al., 2019). Dynamic light scattering (DLS) analysis revealed that more than 90% of the cpPD complexes are approximately 130nm in diameter when in solution (Figure 1C) with a Z-potential of −55mV (Figure 1D). Therefore, the size of these cpPD complexes was three times smaller than in our earlier publication and their Z-potential was halved compared to complexes generated using our previous method (Yoshizumi et al., 2018).

To track the progress of cpPD complexes inside cells, we labeled the plasmid DNA with the fluorescent dye Cy3 and detected the resulting cpPD-Cy3 complexes inside the cell by CLSM (Figure 1E). The cpPD-Cy3 complexes were found as particles inside Arabidopsis cells (Figure 1E). We hypothesize that the larger cpPD-Cy3 complex in planta relative to that obtained by DLS may reflect the spatial resolution limits of CLSM.

Next, we characterized cpPD complexes by FE-SEM. When embedded in 2% agarose gel, cpPD complexes appeared as white globular structures with a black circumference (Figure 2A). To confirm these characteristics in vivo, we infiltrated a solution of cpPD complexes into Arabidopsis leaves and visualized the cpPD complexes in the cells by FE-SEM. We observed a similar appearance of the cpPD complexes in vivo to that seen in vitro (Figure 2B). Some of the cpPD complexes were located close to chloroplasts (Figure 2C). With the higher resolution of SEM, the diameter of the cpPD complexes was about 100nm (Figure 2). These results suggest that the cpPD complexes maintain their shape and size when infiltrated into living plant cells.

To understand how cpPD complexes penetrate into the cell after infiltration from the leaf surface, we tracked the progress of cpPD-Cy3 complexes in Arabidopsis mesophyll cells (Figure 3). We stained the plasma membrane with the dye FM4-64 (Rigal et al., 2015) to visualize the cell boundary at 1, 3, 6, and 12h after infiltration (Figure 3A) and generated a full z-projection from z-stack images with the help of ImageJ (Schindelin et al., 2012; Supplementary Figure 1). The images revealed that the cpPD-Cy3 complexes gradually penetrate into leaf mesophyll cells in a time-dependent manner (Figure 3A).

To clarify how the cpPD-Cy3 complexes penetrate into leaf mesophyll cells, we scored the number of complexes accumulating outside and inside the cell as a function of time since infiltration (Figure 3B). The number of cpPD-Cy3 complexes inside the cell was the lowest after 1h but increased until 6h (Figure 3B), as evidenced by the gradual rise of the average ratio between the complex numbers inside and outside the cell from about 10% at 1h, 20% at 3h, and reaching 35% at 6h (Figure 3C). However, longer incubation times resulted in a decrease of this ratio to about 20% at 12h (Figure 3C). We examined the effect of wortmannin and brefeldin A (BFA), which inhibit endocytosis and autophagy, on the cpPD-Cy3 complexes penetration into the cell (Supplementary Figure 2). The result showed that degradation of the cpPD-Cy3 complexes was suppressed at 12h. These results suggested that cpPD-Cy3 complexes penetrate into the intercellular space of the leaf through stomata within 1h and gradually invade the cell from 3 to 6h before delivering their DNA cargo to the chloroplast, and degrade in vacuole at 12h. Based on the above results, we focused our observations of cpPD-Cy3 complexes penetrating chloroplasts at 6h after infiltration.

We showed previously that cpPD-Cy3 complexes are trapped on the chloroplast outer membrane and become gradually internalized into the organelle (Yoshizumi et al., 2018). To further determine how cpPD-Cy3 complexes enter the chloroplast, we carefully observed individual cpPD-Cy3 complexes near chloroplasts whose outer membranes were fluorescently-labeled by a fusion protein between the N terminus of CHLOROPLAST UNUSUAL POSITIONING1 (CHUP1) and the GFP. Consistent with our previous results, cpPD-Cy3 complexes showed an association with the chloroplast outer membrane (Figure 4A-1), followed by the attachment of the complexes to a membrane protrusion formed on the chloroplast surface (Figure 4A-2), culminating in complexes being enveloped by the protrusion, as would be observed for phagocytosis (Figure 4A-3).

To ascertain that individual cpPD-Cy3 complexes are internalized into the chloroplast, we then performed time-lapse imaging by CLSM. After cpPD-Cy3 complexes were trapped by a chloroplast protrusion (Figure 4B; Supplementary Movie 1), they gradually appeared to be integrated into the chloroplast before becoming undetectable (Figure 4C, white arrows; Supplementary Movie 1). These results confirmed that cpPD-Cy3 complexes are recognized by the chloroplast outer envelope and then internalized into the organelle.

To test whether cpPD-Cy3 complexes reach chloroplast nucleoids (cp-nucleoids; Powikrowska et al., 2014), we stained chloroplast DNA with SYBR Green I (Dragan et al., 2012) and visualized cp-nucleoids and cpPD-Cy3 complexes simultaneously at 6 and 12h after infiltration. The cp-nucleoids appeared as green foci inside chloroplasts, which were detected based on chlorophyll autofluorescence (shown in blue). Importantly, we observed the co-localization of several cpPD-Cy3 complexes (shown in magenta) and cp-nucleoids (Figures 5A,D,G). We classified the appearance of cpPD-Cy3 complexes and adjacent cp-nucleoids into four stages, namely Phase I (a control: chloroplast without the cpPD-Cy3 complexes) and Phases II–IV (Figures 5, 6). Phase II is characterized by the binding of cpPD-Cy3 complexes to the chloroplast surface, with only a few complexes being close to cp-nucleoids (Figures 5A–C). Phase III was associated with many cpPD-Cy3 complexes gathering along the chloroplast surface (Figures 5D–F). Finally, the strong accumulation of DNA molecules, labeled with Cy3 and released from the cpPD-Cy3 complexes, inside the chloroplast, constituted Phase IV (Figures 5G–I). From the images shown in Figures 5A,D,G, we selected individual representative chloroplasts (Figures 5B,E,H) to measure their fluorescence intensity profiles along a line transecting the organelle. The fluorescence profiles of Phase I-IV clearly showed different localization patterns of cpPD-Cy3 complexes and indicated the gradual co-localization of plasmid DNA with cp-nucleoids (Figures 5C,F,I).

To better characterize how plasmid DNA moves from the chloroplast outer membrane to cp-nucleoids, we scored the number of chloroplasts presenting Phases I–IV at 6 and 12h after infiltration (Figure 6A). After 6h, most chloroplasts lacked detectable cpPD complexes around their periphery, while chloroplast with cpPD-Cy3 complexes were broadly evenly distributed across all Phases (Figure 6A). By 12h after infiltration, far fewer chloroplasts were devoid of cpPD-Cy3 complexes nearby, with Phases II–IV demonstrating a steep rise over 6h. In addition, Phase III and Phase IV stages were more frequent after 12h than the Phase II stage, with ~60% of chloroplasts exhibiting a Phase IV stage (Figure 6A). Based on these results, Phase I–IV stages clearly reflected the time-dependent steps followed by cpPD-Cy3 complexes accumulating inside chloroplasts (Figures 5, 6).

To determine if cpPD-Cy3 complexes came in close proximity to cp-nucleoids, we next scored chloroplasts for co-localization between the fluorescence signals derived from Cy3 (cpPD-Cy3 complexes) and SYBR Green I (cp-nucleoids): about 15% of cpPD-Cy3 complexes did colocalize with cp-nucleoids; this proximity would likely allow the incorporation of the plasmid DNA delivered by the complexes into the chloroplast genome (Figure 6B).

As a final test for the effective incorporation of plasmid DNA, Prrn-aadA-sfGFP-Trps, into the chloroplast genome, we attempted to detect fluorescence from a transgene encoding the superfolder GFP (sfGFP; Pedelacq et al., 2006; Fujii and Kodama, 2015). GFP fluorescence in a portion of chloroplast (Figure 6C-1) and most chloroplasts (Figures 6C-2,3) in a cell was observed. These results suggested that the plasmid DNA was incorporated inside the chloroplast stroma and integrated into the chloroplast genome. Based on our observations, the cpPD complexes were clearly internalized into the chloroplast via engulfment by the chloroplast outer envelope.