To ensure functional activity and facilitate future research, syn-LA was designed with maximum similarity to the original C. reinhardtii mitochondria genome (NC_001638.1). Detailed labels and modification for identification and future updates were introduced (Figure 1A; Table 1). Specifically, stop codons were replaced; the TAA codon of cob, nad4, and nad5 were altered to TGA. A further 19 synonymous substitutions referring to nine amino acids, comprising Thr, Val, Ser, Ala, Phe, Arg, His, Leu, and Cys, were equally introduced. Considering that the original cob gene is located at the left end, we shifted it to the intergenic region of nad4 and nad5, as there are two orthologous arms available in contrast with a single arm at the left end. In addition, aiming to assist with transformant screening and achieve homoplasmic status of syn-LA, an 804 bp paromomycin resistance gene aph8 was inserted to the 1,928–2,731 site of syn-LA. Furthermore, to facilitate assembly and convenient future upgrades, four unique restriction sites (PstI, NotI, XhoI, and PmeI) were inserted into the 1–6, 551–558, 2,732–2,737, and 3,890–3,897 sites, respectively. Finally, a 6,090 bp syn-LA sequence sharing 92% identity with the OMtG left arm was obtained.

Benefiting from technological advances, nucleotide molecule synthesis is now extremely convenient, and numerous biotechnology companies provide precise and high-quality DNA synthesis services. The syn-LA sequence was synthesized in four segments (S1–S4) and cloned into the pGEM-T vector by General Biol Company. Specifically, pPstI-S1-NotI-XhoI-SgrDI-SacI, pNotI-S2-XhoI, pXhoI-S3-SgrDI, and pSgrDI-S4-SacI four elementary plasmids were obtained. S2, S3, and S4 fragments were then sequentially cut off from the opposite plasmid through double digests and added to pPstI-S1-NotI-XhoI-SgrDI-SacI in turn (Figure 1B). Thus, final pPstI-S1-NotI-S2-XhoI-S3-SgrDI-S4-SacI containing full-length syn-LA was obtained. To further confirm the assembled syn-LA genomes, PCR amplification was performed (Figure 1C). As primers were designed at positions with insertions or substitutions, the OMtG template could not be bound or extended, making it distinguishable from the synthetic left arm (syn-LA; Supplementary Table 1).

Mitochondria mutant strain cc-2654, which does not grow on acetate plates in darkness, was selected as the recipient strain (Randolph-Anderson et al., 1993; Vahrenholz et al., 1993). The lack of ability of cc-2654 can be restored by complementing the intact mitochondrial cob gene and hence facilitating transformant screening. In consideration of the low efficiency of the existing biolistic mitochondria transformation method, we adopted a modified method. Microcarriers used in mainstream biolistic equipment were restricted to the micron level. The smallest commodity microcarrier is a particle of 0.6 μm gold (Au), which larger than mitochondria (0.5–1 μm). Such particles often cause severe physical damage of the mitochondria when bombarded. Irreversible mitochondrial damage greatly reduces the possibility of successful transfection (Kleele et al., 2021). Thus, a mixture of 0.6 μm gold and 50 nm tungsten (W) powder were applied as our DNA microcarrier to reduce physical damage and prevent aggregation (Figure 2).

Using the modified method, we successfully obtained hundreds of candidate algal colonies after 90 days of dark screening (Figure 2 and Supplementary Figures 1 A,B). To further confirm the presence of syn-LA in those colonies, PCR amplification with two pairs of primers, p1&p2 and p1&p3, was conducted. Primers p1 and p2 are located on aph8 and cob of syn-LA, while primer p3 is located outside syn-LA on cox1 ORF. PCR fragments generated from p1&p2 were as expected, i.e., identical to positive controls. Hence, the results indicated that syn-LA was successfully delivered into cc-2654 cells. The expected 4,499 bp fragment generated from primer pair p1&p3 suggested the success replacement of syn-LA at the corresponding region of OMtG (Figures 3A,B). In addition, some unexpected PCR products were also observed from amplification results of p1&p3, noticed possibly because both the syn-LA and OMtG genome exist in the cell.

To further confirm whether syn-LA was indeed functionalized inside algal cells, mitochondrial proteins were extracted from cc-2654, cc-4098 (allelic mutant of cc-2654), wild-type cc-124 strains, and three syn-LA transformants (LA-T1, LA-T2, and LA-T3) to execute western blotting. The quality of extracted protein was evidenced by the presence of expected 50 and 32 kDa proteins for all six samples after immunoblotting with anti-α-tubulin and anti-VDAC/porin antibodies (Figure 3C). Meanwhile, the 43 kDa protein encoded by cob gene, which should be missing in the cc-2654 and cc-4098 sample, was detected in LA-T1, LA-T2, and LA-T3. Therefore, cob in syn-LA was successfully expressed, indicating that syn-LA should be functionalized in vivo.

As there are usually hundreds or even thousands of mitochondria DNA molecules present in each cell, an occasional mutation or transgenic event is unlikely to alter all mitochondria DNA copies. Thus, the mitochondrial genome can present in a heteroplasmic state within a single cell (Machado et al., 2015; Filograna et al., 2019). Considering the excess bands detected by p1&p3, we also wanted to evaluate the ratio of syn-LA genome in transformants based on quantitative real-time PCR. The single copy nuclear gene CPLD24 (Cre10.g435850) was selected as the internal reference. Total gDNA (genomic DNA) samples from LA-T1, LA-T2, and LA-T3 were extracted and analyzed. Primers targeting aph8 specifically located on syn-LA and targeting nad5 located in both syn-LA and OMtG were designed. In a comparison of the amount of aph8 relative to nad5 in a tested gDNA, the ratios of syn-LA were 0.236, 0.251, and 0.297 in the three transformants LA-T1, LA-T2, and LA-T3, respectively (Figure 3D). This implies that syn-LA covered about one-third of total mitochondrial DNA copies.

In addition, phenotypes of transformants, considering growth status in dark conditions and the reproductive rate under the concurrent condition, were evaluated to estimate the performance of syn-LA in living cells (Figures 3E,F). The results showed that syn-LA-containing algal cells (LA-T1, LA-T2, and LA-T3) adapted better to dark conditions compared to non-transformed cc-2654 and cc-4098. However, it did not perform as well as wild type cc-124. Reproductive rate analysis showed that LA-T1, LA-T2, and LA-T3 grew faster than cc-2654 and cc-4098 but grew slower than cc-124. This result further indicates that syn-LA functions in vivo partially rescue the defect growth of cc-2654 in dark conditions, which was consistent with previous findings that syn-LA only occupied about 30% of total mitochondrial DNA. It was assumed that if syn-LA could replace all OMtG, the defects of cc-2654 could be fully recovered. Consequently, attempts to increase the syn-LA ratio in transformants were necessary.

Many studies have suggested that homoplasmic mitochondrial mutations have milder clinical symptoms than heteroplasmic (Finsterer and Mehri, 2022). Thus, the ability to promote the heteroplasmic mutant to homoplasmic would be useful. In this study, resistance gene aph8 and corresponding paromomycin stress were introduced to increase syn-LA copies in transformants. Certainly, a test to verify aph8 expression is important before doing this. Thus, reverse transcription PCR amplification was conducted with p1&p4, taking rrnL7 (p5&p6) as internal control (Figure 4A and Supplementary Table 1). Results showed that the aph8 gene was successfully transcripted. Further, positive transformants were inoculated onto plates containing 0, 8, 10, 12, 14, and 16 μg/ml paromomycin (Supplementary Figure 1C). It was shown that these transformants could grow on plates containing ≤12 μg/ml paromomycin. Surprisingly, long-term antibiotic stress seemed unlikely to increase the syn-LA ratio. Syn-LA heteroplasmic ratios of samples grown on a plate containing 10 μg/ml paromomycin were at the third, seventh, fifteenth, and thirteeth days (Figure 4B). The result showed that no significant increase of syn-LA heteroplasmic ratio can be observed. We hypothesized that mitochondrial genes located in closed compartments might not be as responsive to environmental stress, contrary to what was originally thought.

In addition, we also found that the mitochondria transformants seem to have the ability to exclude transformed DNA. A test to continuously monitor the heteroplasmic ratio showed syn-LA will gradually reduce and disappear within a month in transformants that grow in constant light conditions (Figure 4C), while transformants cultured in darkness had a relative stable syn-LA ratio. This result may be because syn-LA provides the full cob function, which is necessary for growth in the dark. This result is not surprising, as mitochondria are reported to normally prevent the entrance of foreign nucleic acids. In natural conditions, apart from very few tRNAs, nucleic acid molecules of other forms, whether DNA or RNA, have difficulty crossing this bound organelle (Mileshina et al., 2011; Shikha et al., 2020; Chandel, 2021).

Mitochondrial genome synthesis and intracellular functionalization are undoubtedly meaningful endeavors, whether in the consideration of custom- engineered cell “powerhouses” or treating mitochondrial diseases. However, progress in this field has been slow, and there are barriers that need to be understood. In this study, we assembled a syn-LA of the mitochondrial genome, transferred it into living algal mitochondria, and analyzed the in vivo function of the artificial genome. Based on the difficulties we encountered, we propose the following barriers to be considered and discuss optional improvements within this field.

Due to advanced genome sequencing data and established synthesis technologies, the design and assembly of the mitochondrial genome is relatively straightforward in species with simple mitochondrial genome structures. However, the in vivo function of synthetic genome could be challenging. As shown in Figure 5, synthetic DNA needs to penetrate the cell wall, cytomembrane, and outer and inner membranes of mitochondria—a total of four layers of physical barriers—to get to the right place to be recognized by the mitochondrial transcription and translation machinery. The natural pores of these barriers may be just a few nanometers in size (Schwab et al., 2015). With conventional DNA delivery systems, it is almost impossible to accomplish such a task. The only biolistic delivery system that is in operation does not exhibit satisfying transformation efficiency. Establishing an efficient delivery system is therefore the first challenge. A recent study showed that custom-coated mitochondria-oriented carbon nanotube transmission systems could be an excellent option (Law et al., 2022). The development of carbon nanotube technology has made sizes of 5–20 nm possible. Through coating with mitochondria-directed peptide and conjugation of foreign DNA, nanotubes can be ideal cargo carriers.

The huge number of mitochondria copies is the second challenge. Foreign DNA entry inevitably leads to coexistence of both. It is difficult to artificially increase the proportion of target genome types to achieve homoplasmic status. Long-term stress selection might be an option, but it did not work well in our study. Recent research on mitochondrion editing suggests that mitochondria-directed transcription activator-like effector nuclease may be a feasible choice (Kazama et al., 2019; Arimura et al., 2020; Forner et al., 2022). A stable hereditary cassette with encoding mito-TALEN tools inserted into the nuclear genome could specifically shear OMtG, regardless of the fact that the synthetic genome may contribute to achieving the relevant homoplasmic status (Figure 5).

Additionally, the lack of screening markers and sorting methods are further limitations. Compared with nuclear transformation screening, efficient screening methods such as resistance screening and fluorescence screening have not been established. The only reliable screening approach with mutant rescue is relatively inefficient and time-consuming. In our study, lengthening the screening period to 3 months was needed. Manual and PCR-based sorting methods further reduced the efficiency and limited the scale. Developing an effective screening and high-flux-sorting system is therefore necessary. One study regarding an intelligent image-activated sorting system based on mitochondrial microscopic morphology differences is interesting in this regard (Harmon et al., 2022). Although there may be other obstacles, the three mentioned above are likely to be the main challenges.

A 15,758 bp C. reinhardtii mitochondria genome of C. reinhardtii was obtained from the NCBI database.¹ All designs for syn-LA including deletions, substitutions, and insertions were carried out using Snapgene 4.1.8.² Insert resistance genes (aph8) were codon-optimized based on maximum codon frequency using CodonW _1_4_2 software.

The C. reinhardtii respiratory deficient strain cc-2654 (dum-1, mt⁻) and its alleles cc-2255 and cc-4098 were obtained from the Chlamydomonas Resource Center, University of Minnesota, United States. Wild-type strain cc-124 was from our laboratory algae bank. For data collection, all samples were grown in an illumination incubator on Trisacetate-phosphate (TAP) medium cultured at 25°C with shaking at 150 rpm under constant light (40 μmol photons/m²/s). All experiments were performed with cells in exponential growth phase. Cells were harvested by centrifugation at 3,000 rpm for 5 min and washed twice with distilled water.

Plasmid DNA was extracted using a Plasmid MiniPrep Kit (Transgen, EM101), and genomic DNA was extracted with a Genomic DNA Kit (Transgen, EE101) following the manufacturer’s instructions. DNA quality was assessed using a NanoDrop2000 Ultra Microscope Photometer (Thermo, United States). PCR amplification was performed with Taq DNA Polymerase (Transgen, AP101) based on a standard three-step program (94°C 3 min; 30 cycles of 94°C 5 s, 60°C 30 s, 72°C 10 s/kb, 72°C 5 min).

Transformation was performed as described previously (Hu et al., 2011, 2012) with some modifications. Briefly, exponential growth phase (5–6 × 10⁶ cells ml⁻¹) algae cells were collected by centrifugation and spread on the center of a plate containing TAP medium and cultured overnight. A mixture of 0.6 μm gold and 50 nm tungsten particles (10 + 10 mg) was precipitated in a 1.5 ml Eppendorf tube. The pellet was washed with 70% ethanol and sterile water for 20 min, respectively, with strong oscillation. The mixture was collected by centrifuging and stored in 500 μl 50% glycerol. Before bombardment, 50 μl of the stored mixture of particle carriers, 5 μg DNA, 50 μl 2.5 M CaCl₂, and 20 μl spermidine were added to a 1.5 ml tube, vortexed, and centrifuged. The pellet was washed with 70% ethanol twice, and 200–250 μl pure ethanol was added and vortexed. A 20 μl volume of the mixture was placed on a membrane carrier, air-dried for 10 min, loaded into the biolistic apparatus (Bio-Rad, Hercules, CA, United States), and bombarded at 1,550 psi with helium. Cells were washed and maintained on the TAP medium in darkness for 24 h, then spread on a TAP plate and incubated in darkness for 3 months until green cells had disappeared completely. Subsequently, colonies could be seen after about 7 days of light restoration.

Five liters of exponential growth phase algal cells were harvested by centrifugation at 5,000 g for 5 min, washed twice with distilled water, and resuspended in 5–10 ml buffer A [0.6 M mannitol, 10 mM Tris–HCl, 5 mM EDTA, 0.2% w/v (0.2 g/100 ml) Bovine serum albumin (BSA), and 0.3% w/v polyvinylpyrrolidone 40 (pH = 7.8)]. Intracellular constituents were released by sonication on ice at 240 W for 15 min with 5 s pulses and 5 s pauses. Low-speed centrifugation at 1,000 g and 4°C was performed twice to remove large debris. The supernatant was centrifuged at 15,000 g and 4°C, yielding raw mitochondria.

Proteins of mitochondria were then released in lysis buffer [8 mM Urea, 1% SDS, 40 mM Tris-base, pH = 8.5, 10 mM Dithiothreitol (DTT), and 2 mM Ethylene Diamine Tetraacetic Acid (EDTA) by sonication at 300 W for 20 min with 10 s pulses and 10 s pauses]. Lysates were centrifuged at 15,000 g for 10 min at 4°C, and the supernatant was collected and quantified with a Protein Quantitative Kit (Transgen, DQ111). A 20 μg sample of protein was mixed with SDS-PAGE loading buffer, boiled for 5 min, separated by 4–12% SDS-PAGE, and transferred onto a PVDF (polyvinylidene difluoride) membrane (0.2 μm; Millipore, Massachusetts, United States). Membranes were incubated with blocking buffer for 2 h then incubated with primary antibodies for 9 h at 4°C. Membranes were then incubated with secondary antibody for 2 h at room temperature after washing with TBST three times for 10 min each. Finally, membranes were incubated in enhanced chemiluminescence (ECL) substrate buffer at room temperature for 3 min and visualized by a Tanon-5200 Chemiluminescent Imaging system (Tanon Science and Technology, China).

The mitochondrial heteroplasmy ratio is a statistic-based real-time PCR, according to published research (Machado et al., 2015). Briefly, genomic DNA was extracted with a Genomic DNA Kit (Transgen, EE101). A NanoDrop2000 Ultra Microscope Photometer device was used to assess sample quality and set a unified concentration of 50 ng/μl. The CPLD24 (Cre10.g435850) nuclear gene with only a single copy (Predict online: https://busco.ezlab.org/) was selected as a reference gene, and H01&H02 primers were designed. H03&H04 and H05&H06 primer pairs were used to amplify aph8 and nad5. Real-time PCR with the DNA samples prepared above was performed using a qPCR SYBR Green Kit (Yeasen, 11143ES50) on an ABI QuantStudio 6 Flex Detection Device as recommended by the manual. Both aph8 and nad5 were normalized against CPLD24 according to the 2^−ΔΔCt method. Calculating the relative amount _(aph8)/relative amount _(nad5) gave the heteroplasmy ratio of syn-LA as aph8 specifically located on syn-LA, while nad5 is located on both.

For statistical analysis, t-tests were conducted (^*0.01 < p < 0.05, ^**0.001 < p < 0.01, and ^***p < 0.001). Results are mean values ± SD.