Electrophoresis was run at 80 V and then the gel was visualized with a UVP High Performance Ultraviolet Transilluminator . Genomic DNA was also amplified with primer pairs pSA3_Cm_F/pSA3_CmR for the chloramphenicol resistance gene of plasmid pSA3 and gfp_qPCR_F/gfp_qPCR_R for the egfp gene of plasmid pIGSAF. Primers nifH_qPCR_F/nifH_qPCR_R for the nifH nitrogenase gene from the F. alni genome were used as a positive control for the quality of DNA extracts. Additionally, purified plasmid pDiGc from E. coli DH5α was used as a DNA template for egfp amplification as a positive control. PCR products were separated on an agarose gel and extracted as in Plasmid Synthesis, above, and sequenced by the UCDNA Sequencing Facility . Forward and reverse sequences were combined to make fulllength amplicon sequences. These were aligned with MUSCLE with published reference sequences for the egfp and camR genes. Additionally, the sequences were compared by BLAST against gene sequences obtained from the original plasmids . Confocal microscopy was used to visualize expression of egfp in F. alni hyphae and vesicles. egfp was expressed both constitutively, from plasmid pIGSAF, and differentially between N and N media under control of the F. alni nif promoter using plasmid pIGSAFnif. Cultures were imaged after on1 week of incubation from the previous transfer, 25 weeks after transformation, as described in Plasmid Maintenance on Selective Media, below. For imaging, F. alni cultures were grown either in N or N BAPP medium as in Culture Conditions, above,blueberry grow pot and immobilized on glass slides with a drop of 3% molten agarose solution maintained in a water bath at 50◦C. Slides were pre-heated on a slide warmer at 50◦C. Fifteen microliter of Frankia hyphal suspension was then pipetted onto each slide and covered with 35 µl of 3% molten agarose.

A #1.5 coverslip was added to the Frankia cells in agarose, and then the slide was allowed to cool to room temperature. The Frankia preparations were visualized on a Leica TCS SP8 STED 3X confocal microscope with either a 20X objective, or a 100X oil-immersion objective, using either bright field or fluorescence with a HyD detector, in the Advanced Imaging Facility, University of California, Davis. For fluorescence imaging, samples were excited with 488 nm light. The emission wavelengths were collected from 500 to 550 nm. Images were stored as .lif files from Leica LAS X and then viewed in FIJI . For imaging of individual hyphae and vesicles, the 100X objective lens was used and images were taken in Z-stacks with a step size of 0.1 µm. To visualize F. alni colonies at low magnification, Z-stack images were taken with the 20X objective in five steps of 2 µm each and then combined using the highest fluorescent intensity of each pixel . Three separate transformation experiments were carried out, from August 2017 through March 2019, as outlined in Table 2. Transformed cultures were maintained with weekly subculturing into fresh selective media and used in experiments over the course of several months following each transformation. Initial transformations of F. alni with plasmid pIGSAF were performed in August of 2017; the cultures were maintained by subculturing and were visualized by confocal microscopy from December 2017 through March 2018 . Transformation of F. alni with plasmid pIGSAFnif was performed in January 2018, selected as above, and imaged by confocal microscopy in April 2018 . These cultures were maintained by subculturing in selective media for RNA extraction for qPCR through July 2018 . These cultures were transformed in September 2018 and subcultured weekly in selective media through March 2019. The presence of the plasmid was confirmed by DNA extraction and PCR amplification of the egfp gene in March 2019 as described in Frankia Transformation, above.

To test the persistence of plasmid pIGSAF in transformed F. alni without selection, a time-course of growth in media without antibiotics was performed. Fresh cultures were transformed with plasmid pIGSAF and initially selected as described in Frankia Transformation, above. After this 4-week selection process, cultures were then grown in non-selective media: cells were first pelleted and re-suspended in fresh BAPP media, then subcultured without the addition of chloramphenicol into six-well plates and incubated for 1 week. A separate set of cultures from the same stock was maintained in selective media and subcultured at the same time points as a control. Each week both sets of cultures were pelleted and re-suspended in 500 µl fresh BAPP media. The suspensions were homogenized by passage through a 21G needle twice. 250 µl of each homogenate was transferred to 4 ml fresh BAPP media without chloramphenicol in each well and the remaining 250 µl was used for total genomic DNA extraction as described in DNA Extraction above. This process was repeated once per week for 4 weeks. At each sampling point, the relative amount of plasmid in each sample was quantified by qPCR, performed in duplicate for each of three biological replicates, using egfp primers GFP_qPCR_F and GFP_qPCR_R . Fold-change of plasmid between each time point was calculated using the 11 Ct method . The infC gene , amplified from the same DNA extracts with primers infC_qPCR_F and infC_qPCR_R , was used as a control to normalize the amount of DNA in each sample. To determine significant changes in plasmid abundance, twotailed Welch’s t-tests were performed in R on normalized 1 Ct values . Cultures grown with and without selection for 4 weeks were also imaged with fluorescence according to methods in Confocal Microscopy, above. After electroporation F. alni cells formed visible hyphae in culture after about 10 days. When subcultured into chloramphenicolselective media, F. alni cultures transformed with unmethylated plasmid pIGSAF were able to grow, whereas untransformed cultures and those transformed with methylated plasmid were not. When analyzed on the gel, purified plasmid pIGSAF from E. coli formed bands representing linear , circular , and supercoiled plasmid .

Presence of plasmid pIGSAF in transformed cultures was verified by a band present in DNA extracts corresponding to the linear form of plasmid pIGSAF that was not present in extracts from wild-type F. alni . PCR amplification and sequencing of the egfp gene from DNA extracts of transformed F. alni cultures confirmed the identity of the plasmid . Absolute copy number quantification with qPCR estimated the plasmid was present at 12.5 copies of plasmid per molecule of genomic DNA; relative quantification gave a similar estimate of 11.8 copies per genome . Each experiment was performed on cultures maintained with routine weekly subculture for 15–32 weeks post-transformation, as described in the Methods. When imaged under 488 nm wavelength of excitation in the confocal microscope, green fluorescence typical of GFP was observed in hyphae as shown in Figure 4. Wild-type F. alni hyphae displayed autofluorescence around 575 nm as observed by Hahn et al. , but no autofluorescence was observed in the 500–550 nm range . When grown in N culture, transformants carrying the pIGSAFnif plasmid showed significant up-regulation of the egfp gene conjugated to the nif cluster promoter, approximately 100- fold relative to N culture, or approximately 8.5-fold per copy of plasmid . The F. alni nifH gene , used as a positive control for nitrogen fixation, was similarly significantly up-regulated approximately 8.5-fold in N media compared with N cultures. Expression of the F. alni rpoD housekeeping gene , used as a negative control, was not significantly different between N and N cultures . F. alni containing pIGSAFnif grown in N media fluoresced predominantly in the vesicles,square plastic pot observed at 100X magnification . Little to no fluorescence was observed in hyphae. Fluorescence in the vesicles was present both in the spherical portion as well as in the stalk connecting the vesicle to the hyphae. No fluorescence was observed in hyphae or vesicles of wild-type F. alni grown in N media . Due to the step size of 0.1 µm bright fluorescence was only observed when vesicles were in the plane of focus . Few vesicles were present in images likely since the cells were observed after 1 week of culture in N medium. We have shown that F. alni can be stably transformed with an unmethylated replicating plasmid introduced by electroporation. The methods presented here circumvent two major transformation barriers in Frankia. First, the lack of methylation avoids restriction of the plasmid by type IV methyl-directed restriction enzymes. Second, the use of a plasmid replicated and maintained outside the genome does not rely on the reduced homologous recombination rate in actinobacteria . Plasmids pIGSAF and pIGSAFnif were stably maintained in F. alni culture and used to perform routine experiments. Three independent transformations were performed over the course of 2 years and the resulting transformants were maintained on selective media by repeated subculture for at least 7 months .

The presence and stability of the plasmids were confirmed with gels of whole genome DNA extracts , PCR and qPCR amplification of plasmid-bound genes , and visualization of GFP expressed from plasmids . The copy number of plasmid pIGSAF per genome in F. alni was determined to be about 12 copies per cell , in line with findings for other plasmids derived from plasmid pIP501 that have been estimated to be maintained at approximately 10 copies per cell . In addition, plasmid pIGSAF was determined to be stable in non-selective media for a period of at least 3 weeks . Of the restriction enzymes identified in Frankia genomes our analysis indicated that the type IV enzymes posed the most likely barrier to transformation. Types I and II enzymes recognize specific sequence motifs of generally six to eight nucleotides and hence are statistically highly unlikely to be a broad-range barrier to transformation or horizontal transfer . Additionally, in the genome of F. alni ACN14a the majority of types I and II genes showed very low expression in both N-culture and symbiosis . A type IV homolog of an mrr type methyladenine-targeting restriction gene was highly expressed in F. alni in culture , suggesting that DNA with methylated adenine bases is degraded in this organism. Actinobacteria, especially Frankia, express type IV methyl-directed restriction enzyme genes more highly in culture than do proteobacteria and firmicutes , a finding that correlates with previous reports of higher transformation efficiencies with unmethylated plasmids than methylated in Corynebacterium and Streptomyces spp. . Genomes of the majority of actinobacteria are missing homologs of the dam methyltransferase gene whose product is used to mark parent DNA strands during replication, and mutS and mutL that form a complex for the removal and repair of mismatched bases on the daughter strand determined by the methylation of adenine residues . Together, these factors suggest a preference for unmethylated over methylated DNA among most of the actinobacteria. Type IV restriction enzymes have been suggested to have evolved as a counter to phage methylation systems that themselves evolved to evade host restriction systems through the methylation of restriction target sites . Phage genomes adopt the methylation patterns of their previous host thus increasing the likelihood of digestion by actinobacterial enzymes if replicated in a dam + host. The expression of type IV restriction enzymes in actinobacteria therefore could represent an adaptation to prevent infection by phages based on the methylation state of their genomes. Differences in methylation patterns between actinobacteria and other bacterial phyla potentially constitute a barrier to horizontal gene transfer between these groups, including phage-mediated gene transfer. Of particular interest to the evolution of root nodule symbioses is the possibility of transfer of relevant genes between Frankia and the rhizobia, and vice versa. It has been suggested that the nodA gene involved in Nod factor biosynthesis evolved in the actinobacteria, including some Frankia, and was then horizontally transferred to the rhizobia . If type IV restriction enzymes create a barrier to horizontal transfer into actinobacteria from dam + bacteria including proteobacteria, it would seem that horizontal transfer from actinobacteria to other phyla would be more likely than the reverse. However, F. alni was observed to down-regulate its type IV mrr gene substantially in symbiosis . As roots contain much lower concentrations of bacteriophage than the surrounding soil this could represent a decreased necessity for restriction enzymes as a defense mechanism during symbiosis. A potential side-effect of this down-regulation, however, is that the barrier to horizontal transfer posed by type IV enzymes is likely lowered during symbiosis. In plants, the endophytic compartment is dominated by actinobacteria, with specific taxa of other phyla including proteobacteria and bacteroidetes .