Cell-free biogenesis of bacterial division proto-rings that can constrict liposomes

A major challenge towards the realization of an autonomous synthetic cell resides in the encoding of a division machinery in a genetic programme. In the bacterial cell cycle, the assembly of cytoskeletal proteins into a ring defines the division site. At the onset of the formation of the Escherichia coli divisome, a proto-ring consisting of FtsZ and its membrane-recruiting proteins takes place. Here, we show that FtsA-FtsZ ring-like structures driven by cell-free gene expression can be reconstituted on planar membranes and inside liposome compartments. Such cytoskeletal structures are found to constrict the liposome, generating elongated membrane necks and budding vesicles. Additional expression of the FtsZ cross-linker protein ZapA yields more rigid FtsZ bundles that attach to the membrane but fail to produce budding spots or necks in liposomes. These results demonstrate that gene-directed protein synthesis and assembly of membrane-constricting FtsZ-rings can be combined in a liposome-based artificial cell. Godino et al. show that FtsA-FtsZ ring-like structures driven by cell-free gene expression can be reconstituted on planar membranes and inside liposome compartments. These cytoskeletal structures constrict the liposome, generating elongated membrane necks and budding vesicles. This study represents a step forward to realizing genetic programming of synthetic cell division.

C ell-free biology aims at understanding cellular processes by reconstituting biological functions from their isolated elementary components in in vitro model systems. Owing to their openness and easy manipulation, cytoplasmic extracts and systems reconstituted from purified elements are more amenable to customized experimental design and quantitative description compared to living cells. Therefore, the minimal requirements to achieve a particular function can be assessed more reliably. Many complex biological structures and processes taking place in bacterial or eukaryotic cells have already been reconstituted in vitro. Notable achievements include the reconstitution of the minimal translation machinery from Escherichia coli 1 , the yeast DNA replication apparatus 2 , filopodial structures 3 , cytoskeleton self-organization and centrosome positioning 4 , egg cytokinesis signaling 5 , DNA segregation with Par 6 , and clathrincoated buds 7 . Encouraged by the many cellular pieces that have already been reconstituted in vitro, synthetic biologists have now engaged in the construction of an entire cell [8][9][10][11][12] .
One of the hallmarks of living systems is their ability to divide. An obvious starting point to conceiving a biology-inspired division mechanism in artificial cells is to consider the canonical pathways taking place in prokaryotes. In most bacteria, symmetrical cell division proceeds by forming a constriction ring that eventually splits the mother cell into equally sized daughter cells 13 . At an earlier stage of cytokinesis, a proto-ring composed of the FtsZ, FtsA, and ZipA proteins, assembles on the inner leaflet of the cytoplasmic membrane at the future division site [14][15][16] . The tubulin-related FtsZ is the core constituent of the proto-ring. FtsZ is a GTPase that can polymerize into protofilaments 17,18 . Anchoring of FtsZ protofilaments to the cytoplasmic membrane is mediated by ZipA and the actin homolog FtsA [19][20][21] . This process is regulated by accessory proteins belonging to the Zap family 22 .
Earlier attempts to divide cell-like liposomal compartments have focused on the reconstitution of the Z-ring from purified proteins 23,24 . These studies have shown that FtsZ aided by one of its anchoring protein partners-or a chimeric FtsZ bearing a membrane targeting segment 25 -can self-organize into filament patterns on supported lipid membranes [25][26][27][28] . When encapsulated inside vesicles, the elementary cytoskeletal proteins form ring-like structures that can deform the liposome membrane 24,29,30 . Whether FtsZ filaments alone exert a contractile force and contribute to the final stage of division remains a subject of debate 31,32 and evidence for complete liposome division is still lacking.
A conceptual issue that is inherent to reconstitution assays solely relying on purified proteins, is the impossibility to maintain steady amounts of cytoskeletal proteins from internal mechanisms as the compartment undergoes division. Another problem raised by conventional cell-free assays is the use of oversimplified buffer compositions that have been tailored for a particular set of enzymatic reactions but fail to reproduce the cytoplasmic environment.
Herein, we address these issues by encoding E. coli division proteins on DNA templates. Genetic control over protein production offers a general solution to achieve self-replication, as well as self-regulation by establishing feedback loops. In this context, the PURE system, a minimal gene expression system reconstituted primarily from E. coli constituents 1 was employed. Different types of proteins and biological functions have already been synthesized de novo with the PURE system, including membrane-associated proteins [10][11][12]33 . Moreover, by containing all relevant factors for gene expression, the PURE system emulates more closely the molecular composition of the bacterial cytoplasm than simple buffers. In the present study we utilized PUREfrex2.0, which provides the best combination of protein yield and expression lifespan 34,35 .
We show that cell-free expressed FtsA is able to recruit FtsZ polymers, forming large-scale two-dimensional networks of curved and ring-like structures in the absence of bundling factors. When the entire set of reactions is encapsulated inside liposomes, proto-rings of FtsA-FtsZ filaments are found to constrict the vesicle, generating extended membrane necks and budding vesicles, a phenotype that has not been reported before. Coexpression of ZapA, a native stabilizer of FtsZ filaments, yields stiffer FtsZ bundles attached to the membrane that fail to constrict into bud necks. FtsZ cytoskeletal structures are also investigated with ZipA membrane-anchor protein. We find that in our low-volume supported lipid bilayer (SLB) assays with ZipA and ≤3 µM FtsZ, the generic crowding agent Ficoll70 is necessary to elicit bundle formation. Cell-free expressed ZapA obviates the need of Ficoll70 and promotes formation of cytoskeletal networks with different, likely more physiological, morphology, and protein monomer dynamics. The prospects of further improvement suggest that the DNA-programmed hierarchical assembly of the Z-ring in liposomes is a promising strategy for dividing synthetic cells. In addition, our approach to reconstituting cellular processes in PURE system provides a generic platform that fills the gap between classical in vitro and in cellulo experiments.

Results
Cell-free synthesized FtsA drives the formation of curved FtsZ filaments. An essential component of the E. coli division protoring is FtsA, a homolog of actin that anchors FtsZ filaments to the cytoplasmic membrane. To bypass the difficult purification of FtsA 28 , we directly expressed a sequence-optimized ftsA opt gene on an SLB (Fig. 1a). In the presence of 3 µM purified FtsZ-A647, curved filaments and dynamic ring-like structures formed on the membrane (Fig. 1b, Supplementary Fig. 1, Supplementary Note 1, Movie 1), concurring with previous reports 12, 28 .
To obtain quantitative insights about the concentration of cellfree synthesized FtsA, pre-ran PURE system samples were analyzed by liquid chromatography-coupled mass spectrometry (LC-MS) ( Supplementary Fig. 2, Supplementary Table 1). Protein abundance was quantified using an internal standard (QconCAT) for the most C-terminal peptide detected. We deduced that, after 3 h of expression, FtsA concentration on the SLB was 2.2 ± 0.2 μM (mean ± SD, three biological repeats) ( Fig. 1a- 36 . However, overlapping rings and dynamic filaments on a lipid membrane have also been observed at protein ratios similar as in our cell-free assay 26,28 . Note that LC-MS data do not report the concentration of active protein, which may differ from the measured concentration of proteolytic peptides. Promoting lateral interactions of FtsZ protofilaments stimulates the formation of higher-order cytoskeletal structures in vitro 37 . However, little is known about how the nature of these lateral interactions influences the morphology of the FtsZ network. Therefore, we decided to investigate the architecture and dynamics of FtsZ protofilaments in a molecular environment that favors lateral interactions. First, we employed Ficoll70, a generic crowding agent known to elicit FtsZ bundle formation (Fig. 1c). Large SLB areas were covered with curved filaments, rings of different sizes (most having a diameter of 1-2 µm, phenotype 1) and large circular patterns (phenotype 2) (Fig. 1d).
Quantitative image analysis revealed that both the curvature and occurrence of branching points of the filament network were reduced in the presence of either Ficoll70 or ZapA (Supplementary Fig. 3b). Differences between the two phenotypes observed in each condition were also quantified ( Supplementary Fig. 3c). We speculate that local changes, such as protein concentration (e.g., owing to edge effects in the chamber) and possible defects in the membrane, may play a role in the nucleation, morphology, or dynamics of the cytoskeletal network, therefore driving formation of one versus the other phenotype on the same SLB.
We challenged the PURE system to co-express both FtsA and ZapA in a single reaction. Given the low amount of ZapA produced in a single-gene reaction and the extra burden imposed on the biosynthesis machinery when expressing an additional gene, we attempted to boost ZapA concentration by substituting the native gene with a sequence-optimized zapA opt DNA construct ( Supplementary Fig. 4). Co-expression of the ftsA opt and zapA opt genes in a one-pot PURE system reaction led also to the formation of bended and straight bundles (Fig. 1g, h, Supplementary Fig. 5) that qualitatively resemble phenotype 1 obtained in separate expression (Fig. 1e, f). Protein quantification was performed by LC-MS, and concentration values of 2.4 ± 0.3 µM FtsA and 0.4 ± 0.2 µM ZapA in the SLB assay were determined (mean ± SD, three biological repeats, Supplementary Table 3). Quantitative inspection of the protein patterns in coexpression experiments shows a lower curvature and branch point density than in single-gene expression assays (Supplementary Fig. 3b), which may be attributed to higher protein concentrations on the SLB when FtsA and ZapA are coexpressed (Supplementary Table 3). Note again that LC-MS data do not provide an accurate measure of the concentration of active protein. Furthermore, usage of PURE system substrates and cofactors, such as the tRNAs and NTPs, is different in a singlegene or double-gene expression (e.g., nucleotide and codon abundance), which may influence reactions not directly involved in transcription/translation. Despite morphological differences observed in the protein patterns between the Ficoll70-and ZapA-containing samples, quantitative analysis of FtsZ subunit dynamics by fluorescence recovery after photobleaching (FRAP) revealed similar recovery halftime values under the tested conditions ( Supplementary Fig. 6).
Bundling enables long sZipA-FtsZ cytoskeletal structure formation. We then examined the self-organization of FtsZ with the membrane-anchor soluble ZipA (sZipA) using purified proteins supplied in a minimal buffer or in PUREfrex2.0 background. The soluble variant sZipA does not contain the transmembrane region by elimination of the hydrophobic N-terminal domain (amino acids 1-25) of the full-length protein. This domain was substituted by a His 6 -tag for binding NTA-conjugated lipids. The FtsZ-binding properties of sZipA are essentially the same as those of the native protein when incorporated in nanodiscs 41,42 . We found that in our low-volume SLB assays with ≤3 µM FtsZ, filament bundling with either Ficoll70 or ZapA was required to trigger large-scale cytoskeletal networks (Fig. 2a, b, Supplementary Note 2). This result contrasts with previous observations 20, 28 , highlighting the role of the total reaction volume as a control parameter in the formation of cytoskeletal patterns. The network morphology, as well as the sZipA and FtsZ monomer dynamics differ whether lateral interactions are promoted by the artificial molecular crowder Ficoll70 or by cell-free synthesized ZapA ( Fig. 2c-f, Supplementary Figs. 7-11). Moreover, we found that increasing the expression level of ZapA (Supplementary Fig. 4) results in more stable filaments that can even extend above the SLB (Supplementary Figs. 8 and 9). Taken together, our results show that ZapA encourages the formation of membranebound sZipA-FtsZ filament network having a different-presumably more physiological-morphology and subunit turnover compared with Ficoll70.
We conclude from these SLB experiments that short, curved filaments and rings that resemble physiological structures are more prominent with FtsA compared with sZipA, and can develop in the absence of a bundling agent.
FtsZ and internally synthesized FtsA constrict liposomes. The identification of FtsZ and FtsA as the minimal molecular set to obtain membrane-anchored curved filaments and rings in the PURE system prompted us to reconstitute FtsA-FtsZ cytoskeletal networks inside liposomes (Fig. 3a). The cell-free gene expression solution was supplemented with adenosine triphosphate (ATP, additional 2 mM), guanosine triphosphate (GTP, additional 2 mM) and a mixture of highly purified chaperones (DnaK mix). Although energy regeneration components are present in the PURE system, extra ATP and GTP were provided to compensate for the extra demand from FtsA and FtsZ. Purified FtsZ-A647 was used to visualize protein localization by laser scanning confocal microscopy. FtsZ-A647 was employed at 3 µM concentration, which is similar to that measured in vivo (~3.5 µM) 36 . Liposomes were formed by natural swelling, with a composition of zwitterionic PC and PE phospholipids, anionic PG and cardiolipin, and a small fraction of TexasRed-conjugated lipid for membrane imaging 35 . Such a lipid mixture and liposome preparation method have proved compatible with the cell-free synthesis of membrane-associated enzymes 10 , DNA replication proteins 11 and division proteins 12 . Liposome size distribution ranges from~1 µm up to over 15 µm in diameter, which provides a more relevant bacterial cell-size compartment than >20 µm liposomes produced with other methods 23,33 . In contrast with previous studies 33 , no crowding agent was included during liposome formation. In fact, we found that Ficoll70 impairs formation of gene-expressing liposomes with our methodology ( Supplementary Fig. 12).
In control experiments where the ftsA opt gene was omitted, FtsZ was exclusively located in the liposome lumen (Supplementary Fig. 13). De novo synthesized FtsA successfully recruited FtsZ on the membrane as shown by the colocalization of the FtsZ-A647 and membrane dye signals (Fig. 3, Supplementary  Fig. 14). Although homogeneous recruitment of FtsZ to the membrane was commonly found within the liposome population, the majority of the liposomes displayed regions with patches of FtsZ on the inner surface of the membrane (Fig. 3b). Noticeably, the membrane spots with clustered FtsZ coincide with different types of membrane remodeling. In some cases, the recruited FtsZ localizes with outward membrane deformation or short protrusions (Fig. 3c, Supplementary Fig. 15). In other instances, the protrusions developed into vesicles or blebs tethered to the parental liposome through a membrane neck coated with FtsZ ( Fig. 3d, Supplementary Fig. 15). Sometimes, the budding neck extends over a few microns in the form of a tubular structure containing one or a few FtsA-FtsZ rings (Fig. 3d, e). Interestingly, these blebbing structures are dynamic. Events, such as appearance of new constriction sites, growing vesicles and diffusion of protein rings along the tube axis were observed (Fig. 3e, Movie 2). Although membrane recruitment of FtsZ in the form of patches was visible already within 2 h of expression, major liposomeremodeling events, such as budding spots and elongated blebs were observed only after 3-4 h. Moreover, after 6 h expression, small vesicles were found to agglutinate to larger liposomes ( Supplementary Fig. 13). FtsA concentration does not significantly increase beyond the first 3 h of expression ( Supplementary  Fig. 2), in agreement with the kinetic profiles of protein production with the PURE system 34 . Concentration of synthesized FtsA was compared after 3 and 6 h expression, yielding 4.5 ± 0.5 μM and 5.8 ± 1.1 μM, respectively. Therefore, we do not expect that the differences observed at incubation times longer than 3 h can be attributed to an increase in protein concentration. Instead, the time-dependent changes could be due to some delaying factors, such as recruitment of proteins to the membrane, assembly of filaments and bundles, protein clustering into patches and remodeling of the membrane. It is unclear whether the FtsZ-coated membrane necks can close to release mature vesicles. Therefore, we cannot ascertain that the small vesicles observed after 6 h are reminiscent to division events. Yet, these aggregated vesicles were not observed when the ftsA opt gene was omitted ( Supplementary Fig. 13), indicating that this global remodeling is dependent on the expression of FtsA.
We then decided to investigate how the presence of cell-free expressed ZapA would modulate the properties of the cytoskeletal  Supplementary Fig. 7. e Schematic illustration of the SLB assays with purified FtsZ-A647 (3 µM) and cell-free synthesized ZapA incubated on top of an sZipA-A488-bound SLB. ZapA was expressed from the native gene zapA. f Fluorescence images of sZipA-A488 (left) and FtsZ-A647 (right) in a sample containing cell-free synthesized ZapA and additional 2 mM GTP. Different cytoskeletal network phenotypes were observed when ZapA concentration was increased upon expression of the optimized zapA opt construct ( Supplementary Fig. 9). Scale bars indicate 10 µm. patterns in liposomes (Fig. 4a). Co-expression of ftsA opt and zapA opt DNA constructs induced formation of FtsZ-A647 clusters on the inner surface of the membrane (Fig. 4b). Liposomes with different cytoskeletal protein phenotypes were observed, such as homogeneous coating to the membrane, patches or filaments, and large ring-like structures. Bundles of FtsZ polymers adopting apparent ring-like structures predominantly localize at the interface of two liposomes (Fig. 4b), coinciding with a membrane septum (i.e., a membrane separating two adjacent vesicles; it could be a single bilayer or two bilayers). However, ZapA abolishes the formation of membrane protrusions, vesicle budding and clustering of FtsZ on tubular membrane structures (Fig. 4b). We have seen that, in the presence of ZapA, the small ring-like structures do not form on SLB, where longer, curved filaments dominate (Fig. 1). We observed a decrease of the filament curvature in the presence of ZapA, especially during co-expression of FtsA and ZapA ( Supplementary  Fig. 3b), which correlates with a higher concentration of ZapA in the assay (Supplementary Table 3). The straighter cytoskeletal filaments are likely not able to develop into contractile rings. Instead, they accommodate to the large compartment and are unable to deform the membrane into narrow necks (Fig. 4b). These results indicate that the mechanical properties of FtsZ-ZapA bundles impede the formation of membrane-constricting, highcurvature cytoskeletal filaments, which suggests that temporal Collectively, the results demonstrate that gene-based reconstitution of membrane-constricting cytoskeletal protein filaments within liposomes is feasible. Moreover, FtsA and FtsZ form the minimal architecture to establish E. coli cell division proto-rings from native constituents in vitro.

Discussion
Compartmentalization of PURE system and of the Z-ring constituents inside liposomes provides a realistic cellular environment. Purified FtsZ and FtsA proteins have already been enclosed within small (diameter < 200 nm) liposomes 30 or giant vesicles 43 . In these earlier studies, membrane-tethered protofilaments of FtsZ could assemble with FtsA*, a mutant of FtsA that cannot polymerize 44,45 . In another report, the FtsZ-sfGFP fusion protein was recruited to the membrane of giant liposomes (diameter 15-100 µm) by FtsA in the presence of Ficoll70 33 . In their work, Furusato et al. 33 reported a homogenous recruitment of FtsZ to the membrane in the presence of FtsA but no membrane deformation. Local reshaping of liposomes was exclusively observed with ZipA as the FtsZ membrane anchor, but no constriction sites nor protein ring-like structures were observed 33 . Here, we show that wild-type FtsA and FtsZ are capable to deform the membrane in PURE system-loaded liposomes with a size <15 µm. The FtsA-dependent recruitment of FtsZ on the membrane frequently induces the formation of FtsZ clusters that constrict the liposome membrane into bud necks. It is clear from our data that FtsZ, assisted by FtsA, does not preferentially accommodate to pre-existing membrane areas with a high curvature. Conversely, membrane constriction and extended neck-like regions connecting the mother and budding vesicles are the product of localized FtsZ-FtsA pattern assembly. Noteworthily, these types of membrane remodeling, cytoskeletal protein organization, and dynamic blebbing structures were not observed in previous reports 23,24,33,46 .
Not every liposome exhibits the same phenotype with regard to FtsZ recruitment and membrane deformation. This disparity is presumably the manifestation of the probabilistic encapsulation of all PURE system components and DNA, which leads to a large heterogeneity in FtsA expression levels, as recently quantified with a fluorescence reporter gene 35 . It is therefore difficult to know the precise concentration of synthesized FtsA in individual liposomes and to correlate it with a particular phenotype.
Further investigations will be necessary for directing the assembly of an all-gene-based contractile FtsZ proto-ring that can divide liposomes. Although we do not exclude that assisting proteins, such as the Min system 12 and the FtsZ-interacting partners SlmA 47 and ZapB 48 , might have to be introduced to complete membrane scission and release budding vesicles, the present results suggest that expression of FtsA and FtsZ might suffice to generate daughter vesicles of a few microns in size. The precise timing of protein interaction is essential for the hierarchical assembly of the proto-ring. This represents a major issue that is inherent to in-liposome compartmentalization of purified cytoskeletal proteins or with temporally unregulated expression of multiple genes. An additional level of temporal control that might be decisive for sequential assembly of the Z-ring constituents could be provided by regulating the expression of individual genes through transcriptional circuits, such as cascade or feedback motifs 49 . Mindful of the limitations to apprehend the PURE system 34 and to rationally design liposomes harboring desirable properties encoded in genes, we believe that in vitro evolutionary optimization, by exploring a wide genetic diversity, provides additional opportunities to build cellular functions, and FtsZ proto-rings in particular.

Methods
DNA constructs. ftsZ and ftsA gene fragments were amplified by standard polymerase chain reaction (PCR) from the chromosomal E. coli BL21 DNA with primers 509 and 374 (ftsZ), and 508 and 376 (ftsA) (Supplementary Table 4). These primers contain overhangs for Gibson assembly with the pET11-a plasmid. PCR products were checked on a 1% agarose gel stained with EtBr or SYBR safe, imaged with a ChemiDocTM Imaging System (BioRad Laboratories), and purified with the Wizard SV Gel kit (Promega). The purified DNA was incubated with DpnI (New England BioLabs®, Inc.) to remove residual plasmid and the linear DNA was purified again with Wizard SV Gel kit. DNA concentration and purity were measured using a ND-1000 UV-Vis Spectrophotometer (Nanodrop Technologies). COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-020-01258-9 ARTICLE COMMUNICATIONS BIOLOGY | (2020) 3:539 | https://doi.org/10.1038/s42003-020-01258-9 | www.nature.com/commsbio fresh prechilled liquid lysogeny broth (LB) medium and incubated for 1 h at 37°C and 250 rpm. The cultures were plated on solid LB medium with ampicillin and grew overnight at 37°C. Colonies were picked up and cultured in 1 mL of liquid LB medium with 50 µg µL -1 of ampicillin in 1.5-mL Eppendorf tubes for 6 h at 37°C and 250 rpm. Plasmid purification was performed using the PureYield™ Plasmid Miniprep System (column method, Promega). Plasmid concentration and purity were checked on a Nanodrop. Linear templates for PURE system reactions were prepared by PCR using the plasmids as substrates with primers 194 and 709 (Supplementary Table 4). Amplification products were checked on a 1% agarose gel and were purified using the Wizard SV Gel kit. DNA concentration and purity were measured using a ND-1000 UV-Vis Spectrophotometer (Nanodrop Technologies).
The DNA fragment containing the zapA gene (original sequence from E. coli K12 strain) was inserted in a pIDTSMART-AMP plasmid (Integrated DNA Technologies). The plasmid was transformed into E. coli TOP10 cells. Transformation, plasmid purification, and production of linear DNA templates were performed as described above.
The ftsA opt and zapA opt constructs (starting with a T7 promoter and ending with the T7 terminator) were sequence-optimized for codon usage, GC content and 5′ mRNA secondary structures, and were inserted in a pJET1 and pUC57 plasmid, respectively (GeneScript). Plasmids were amplified and purified as described above. All sequences of the linearized constructs can be found in the Supplementary Methods. Cell-free gene expression. PUREfrex2.0 (GeneFrontier Corporation, Japan) was utilized following storing and handling instructions provided by the supplier. Linear DNA templates were used in single-gene expression assays at a final concentration of 5 nM. In co-expression experiments, both ftsA opt and zapA opt constructs were included at 5 nM and 10 nM, respectively, along with 1 µL of DnaK Mix (GeneFrontier Corporation). DnaK Mix consists of highly purified E. coli DnaK, DnaJ, and GrpE chaperone proteins. Reactions of 20 µL volume were carried out in test tubes for 3 h at 37°C. When indicated, samples were supplemented with purified proteins (FtsZ-A647, sZipA-A488) and added either on top of an SLB or used for lipid film swelling.
QconCAT purification. QconCAT was designed to contain two specific peptides for FtsA and two for ZapA (Supplementary Table 1, Supplementary Fig. 2). QconCAT was expressed in BL21(DE3) cells in M9 medium with 15 NH 4 Cl and ampicillin. A pre-culture was diluted 1:100 to a 50-mL expression culture. Protein expression was induced at OD 600 = 0.5 with 1 mM isopropyl β-d-1thiogalactopyranoside and cells were grown for 3 h at 37°C. Cells were harvested by centrifugation and the pellet was dissolved in 1 mL B-PER. 10 µL of 10 mg mL -1 lysozyme and 10 µL of DNaseI (ThermoScientific, 1 U µL -1 ) were added and the sample was incubated for 10 min at room temperature. The lysate was centrifuged for 20 min at 16,000 × g and the pellet resuspended in 2 mL of a 1:10 dilution of B-PER in MilliQ water. The sample was twice again centrifuged, and the pellet was resuspended in 2 mL 1:10 diluted B-PER and centrifuged again. The pellet was resuspended in 600 µL of 10 mM Tris-HCl pH 8.0, 6 M guanidinium chloride and incubated at room temperature for 30 min. After spinning down unsolubilised protein the supernatant was loaded onto an equilibrated mini NiNTA spin column and the flow-through was reloaded twice to maximize protein binding. The column was washed twice with 600 µL of 10 mM Tris-HCl pH 6. Trypsin digest. Per LC-MS injection, 1.5 µL of PURE system reaction was mixed with 3 µL of 100 mM Tris-HCl pH 8.0, 0.3 µL of 20 mM CaCl 2 , and 0.8 µL MilliQ water. Samples were incubated at 90°C for 10 min to stop the reaction. Then, 0.6 µL of QconCAT (0.3 mg mL -1 ) was added, the sample was incubated again at 90°C for 10 min and after cooling to room temperature 0.3 µL of 1 mg mL -1 trypsin (trypsin-ultra, MS-grade, New England Biolabs) was added. Samples were then incubated at 37°C overnight. After addition of 0.7 µL 10% trifluoroacetic acid, samples were centrifuged in a table-top centrifuge (5415 R, Eppendorf) for 10 min at maximum speed. The supernatant was transferred to a glass vial with smallvolume insert for LC-MS/MS analysis.
LC-MS/MS analysis. LC-MS/MS analysis was performed on a 6460 Triple Quad LCMS system (Agilent Technologies, USA) using Skyline software 52 . In all, 7 µL of sample was injected per run to an ACQUITY UPLC® Peptide CSH™ C18 Column (Waters Corporation, USA). The peptides were separated in a gradient of buffer A (25 mM formic acid in MilliQ water) and buffer B (50 mM formic acid in acetonitrile) at a flow rate of 500 µL per minute and at a column temperature of 40°C. The column was equilibrated with 98% buffer A. After injection, the gradient was changed linearly over 20 min to 70% buffer A, over the next 4 min to 60% buffer A, and over the next 30 s to 20% buffer A. This ratio was held for another 30 s and the column was finally flushed with 98% buffer A to equilibrate for the next run. Selected peptides were measured by multiple reaction monitoring. For both ZapA and FtsA, two peptides were present in the QconCAT. In addition, two peptides from ribosomal proteins were also measured as control.
Labeling of in vitro synthesized proteins and gel analysis. PUREfrex2.0 reaction mixtures were supplemented with 0.5 μL of GreenLys (FluoroTect TM GreenLys, Promega) and gene expression was performed in a test tube as described above. Samples were treated with RNase (RNaseA Solution, Promega) for 30 min and proteins were denatured for 10 min at 90°C in 2× SDS loading buffer with 10 mM DTT. Samples were loaded on a 18% sodium dodecyl sulphate-polyacrylamide gel electrophoresis gel. Visualization of the fluorescently labeled translation products was performed on a fluorescence gel imager (Typhoon, Amersham Biosciences) using a 488-nm laser and a band pass emission filter of 520 nm. Formation of SLBs. The imaging chamber was treated with oxygen plasma (Harrick Plasma basic plasma cleaner) for 30 min to activate the glass surface. Immediately after plasma cleaning the SUV solution was added to the sample reservoir at a final lipid concentration of 0.94 mg mL -1 together with 3 mM CaCl 2 . The chamber was closed by sticking a coverslip using a double-sided adhesive silicone sheet (Life Technologies) and the sample was incubated for 30 min at 37°C. Next, the chamber was opened and the SLB was carefully washed six times with SLB buffer. Under these conditions, the SLB contains 4.8 molar % of 18:1 DGS-NTA (Ni 2+ ) lipids, which is within the range studied by ref. 27 (0.5-10 mol %), similar as in ref. 28 (1-8 mol%) but higher than in ref. 38 (0.02-0.08 mol% of full-length ZipA, DGS-NTA lipid was not used in this study) and lower than in ref. 50 (10 mol%).
Activity assays on supported membranes. In the experiments involving sZipA-A488, 1 µM of the purified protein was first incubated on top of an SLB for 10 min at room temperature. The SLB was washed with 10 µL reaction buffer (50 mM Tris-HCl, 150 mM KCl, 5 mM MgCl 2 , pH 7.5). Then, 20 µL of sample (composition is specified where relevant) was added on top of the SLB and the chamber was sealed by sticking a 20 × 20 mm coverslip with a double-sided adhesive silicone sheet. In the experiments with FtsA, the ftsA or ftsA opt gene was either directly expressed on top of the SLB, or in a test tube and subsequently added onto an SLB as part of the sample. In the earlier configuration, a 20 µL PUREfrex2.0 reaction was carried out on top of an SLB and 10 µL were removed and replaced by the activity assay mixture. The exact composition of the sample varies for the different experiments and is specified where appropriate. In all cases, samples contained 2 mM GTP, supplemented with 2 mM ATP in FtsA experiments. In all assays without ZapA, Ficoll70 was added to a final concentration of 50 g L -1 . No oxygen-scavenging system was used, unlike in ref. 28 but like in ref. 38 .
Spinning disk microscopy. SLBs were imaged with an Olympus iX81 inverted fluorescence microscope equipped with a ×100 oil immersion objective (Olympus), an iXon3 EMCCD camera (Andor Technology) and a Nipkow spinning disk (CSU-XI, Yokogawa). FtsZ-A647 and sZipA-A488 were imaged using a 640 nm and 491 nm laser line, respectively, and appropriate emission filters (685/40 nm or 525/50 nm). The software Andor IQ3 (Andor Technology Ltd.) was used for image acquisition and identical settings were used for all experiments. Experiments were conducted at room temperature.
Preparation of lipid-coated beads. Lipid-coated beads were prepared according to our published protocol 35 with the following lipid composition: DOPC (50 mol %), DOPE (36 mol%), DOPG (12 mol%), 18:1 CL (2 mol%), DSPE-PEG-biotin (1 mass%) and DHPE-TexasRed (0.5 mass%) for a total mass of 2 mg. Lipids dissolved in chloroform were mixed in a round-bottom glass flask. Methanol containing 100 mM rhamnose (Sigma Aldrich) was added to the solution in a chloroform-to-methanol volume ratio of 2.5:1. Then, 1.5 g of 212-300 µm glass beads (acid washed, Sigma Aldrich) were poured to the lipid-rhamnose mixture and the organic solvent was removed by rotary evaporation at 200 mbar for 2 h at room temperature, followed by overnight desiccation. Lipid-coated beads were stored under argon at -20°C until use.
Production and immobilization of gene-expressing liposomes. A PUREfrex2.0 reaction mixture was assembled as described above. Either or both ftsA opt and zapA opt DNA constructs were added at a final concentration of 5 nM and 10 nM, respectively. The solution was supplemented with (final concentrations indicated): 2 mM GTP, 2 mM ATP, 3 µM FtsZ-A647 and MilliQ to reach a final volume of 20 µL. About 20 mg of lipid-coated beads was added to the solution and liposomes were formed by natural swelling of the lipid film for 2 h on ice, protected from light. During incubation, the tube was gently rotated manually a few times. Four freeze-thaw cycles were then applied by dipping the sample in liquid nitrogen and thawing on ice. The sample reservoir of the imaging chamber was functionalized with 1:1 molar ratio of bovine serum albumin (BSA) and BSA-biotin (1 mg mL -1 , ThermoFisher Scientific), and then with Neutravidin (1 mg mL -1 , Sigma Aldrich), to tether the biotinylated liposomes. About 7 μL of the liposome solution was carefully pipetted (with a cut tip) into the imaging chamber and supplemented with RQ1 DNase (0.07 U µL -1 ) to preclude gene expression outside liposomes. The chamber was sealed by sticking a 20 × 20 mm coverslip with a double-sided adhesive silicone sheet. Expression was performed directly on the confocal microscope at 37°C for 1.5-6 h.
Confocal microscopy. A Nikon A1R Laser scanning confocal microscope equipped with an SR Apo TIRF ×100 oil immersion objective was used to image liposomes. The 561 nm and 640 nm laser lines were used in combination with appropriate emission filters to image the Texas Red membrane dye and FtsZ-A647, respectively. The software NIS (Nikon) was used for image acquisition and identical settings were used for all experiments. Samples were mounted on a temperature-controlled stage maintained at 37°C during imaging.
Fluorescence recovery after photobleaching. FRAP experiments were performed on an Olympus iX81 spinning disk microscope. Images were acquired using the following protocol: 10 frames every s, 10 frames every 250 ms, 10 frames every 2 s, 10 frames every 4 s. Analysis of the FRAP images was performed with ImageJ 53,54 using the FRAP profiler plug-in. The intensity of a bleached region of interest (ROI, 29 × 29 pixels) was measured over time and normalized to the intensity of the surrounding (250 × 250 pixels area centered on the ROI) to correct for the bleaching that occurs during image acquisition. Fitting of the FRAP curves was generated in GraphPad Software Inc. using a one-phase exponential model. At least three FRAP measurements were performed in each sample analyzed.
Structured illumination microscopy (SIM). 3D SIM images have been acquired with a Nikon SIM microscope equipped with a Nikon ×100 and 1.49 NA Apo TIRF SR objective and a 640 nm laser line. The acquisition and reconstruction of the SIM images have been performed using the Nikon NIS element software. SIM raw data and their corresponding reconstructed images were quality-checked using the Fiji plug-in SIMcheck 55 .
Total internal reflection fluorescence microscopy. FtsA-FtsZ ring dynamic was investigated using a Nikon TiE inverted fluorescence microscope equipped with an iLAS2 illumination system, a Plan Apo ×100 oil immersion objective and a 2× Photometrics EMCCD Evolve Camera. The 640 nm laser line was used in combination with appropriate emission filters to image FtsZ-A647. MetaMorph® Microscopy Automation Software was used for image acquisition.
Quantitative image analysis. Image analysis was performed using Mathematica (Wolfram Research, version 11.3). All images were corrected for uneven illumination by applying a Gaussian filter with radius 70 pixels to each image, fitting a third-degree polynomial to the filtered image, and dividing the original image pixel-by-pixel with the fitted polynomial. To segment the filaments a ridge filter with sigma = 1 was applied and the resulting image was binarized using morphological binarization with the default parameters. When needed this image was convolved with a Laplacian-of-Gaussian filter with radius two pixels and inversed (necessary for images with thin filaments). Filament thicknesses were calculated from this image as the distance of the centerline of filaments to the edge using the distance transform function. Branch point density and filament curvature were calculated after the thinning operation was applied to the segmented image. Curvatures were approximated at each pixel along the thinned filament with a distance larger than two pixels to the next branch point using Gaussian smoothing on the derivative functions. The image processing steps are illustrated in Supplementary Fig. 3.
Filament dynamic was analyzed by computing kymographs in MatLab version 2018a. A user-defined ellipse was overlaid on the ring-like structures of interest based on the first image of the movie and the boundary pixel intensities of all frames were extracted.