Conserved and repetitive motifs in an intrinsically disordered protein drive ⍺-carboxysome assembly

All cyanobacteria and some chemoautotrophic bacteria fix CO2 into sugars using specialized proteinaceous compartments called carboxysomes. Carboxysomes enclose the enzymes Rubisco and carbonic anhydrase inside a layer of shell proteins to increase the CO2 concentration for efficient carbon fixation by Rubisco. In the ⍺-carboxysome lineage, a disordered and highly repetitive protein named CsoS2 is essential for carboxysome formation and function. Without it, the bacteria require high CO2 to grow. How does a protein predicted to be lacking structure serve as the architectural scaffold for such a vital cellular compartment? In this study, we identify key residues present in the repeats of CsoS2, VTG and Y, which are necessary for building functional ⍺-carboxysomes in vivo. These highly conserved and repetitive residues contribute to the multivalent binding interaction and phase separation behavior between CsoS2 and shell proteins. We also demonstrate 3-component reconstitution of CsoS2, Rubisco, and shell proteins into spherical condensates and show the utility of reconstitution as a biochemical tool to study carboxysome biogenesis. The precise self-assembly of thousands of proteins is crucial for carboxysome formation, and understanding this process could enable their use in alternative biological hosts or industrial processes as effective tools to fix carbon.


Tables:
S1 Strains used in H. neapolitanus growth experiments.

S2
Sequences used in this study.The ∆csoS2 strain is a knockout of CsoS2 via spectinomycin ORF insertion.The CsoS2 complement strain contains a genomic insertion of IPTG-inducible CsoS2 and kanamycin targeted to a neutral site on the genome in the ∆csoS2 background strain.WT, ∆csoS2, and CsoS2 complement strains were plated in a serial dilution and grown in high CO2 (5% CO2) or air (0.04% CO2) on DSMZ68-agar plates + 100 μM IPTG with (A) no antibiotic, (B) 10 μg/ml spectinomycin, and (C) 10 μg/ml spectinomycin + 2 μg/ml kanamycin.Some images are a composite from different rows on the same plate that were not immediately next to each other, as marked by a thin white line.Representative gel (same CsoS2 agarose gel that is in Figure 3B), with boxes drawn around quantified areas.Blue boxes indicate areas of the gel that were quantified to measure shell migration, and yellow boxes indicate areas of the gel that were quantified to measure BSA migration.For shell, density in boxes 2 and 5 were averaged and subtracted as background, since there was no shell present in these samples.
For BSA, density in boxes 1, 2, and 3 were averaged and subtracted as background, since no BSA was present in these samples.Three gels for each sample were run and quantified in this way, which is summarized in Figure 3C.

Figure S1 .
Figure S1.CsoS2 knockouts in H. neapolitanus are partially complemented in air.The ∆csoS2 strain is a knockout of CsoS2 via spectinomycin ORF insertion.The CsoS2 complement strain contains a genomic insertion of IPTG-inducible CsoS2 and kanamycin targeted to a neutral site on the genome in the ∆csoS2 background strain.WT, ∆csoS2, and CsoS2 complement strains were plated in a serial dilution and grown in high CO2 (5% CO2) or air (0.04% CO2) on DSMZ68-agar plates + 100 μM IPTG with (A) no antibiotic, (B) 10 μg/ml spectinomycin, and (C) 10 μg/ml spectinomycin + 2 μg/ml kanamycin.Some images are a composite from different rows on the same plate that were not immediately next to each other, as marked by a thin white line.

Figure S2 .
Figure S2.Illustration of all CsoS2 variants expressed in vivo in H. neapolitanus.Note for the C→S variants, there are no cysteines in R3.Note for the VTG variants, only VTG and VSG sites were mutated.

Figure S3 .
Figure S3.Western blots of complemented CsoS2 and mutant strains.Strains were grown in liquid culture at high CO2 (5%) +/-100 μM IPTG and appropriate antibiotics.(A) C→S and VTG→AAA mutants blotted with antibody 1; loading was normalized to ~25 μg/sample.Some antibodies do not bind due to the mutated VTG sequence in the strain.(B) VTG→AAA mutants blotted with antibody 2; loading was normalized to 50 μg/sample.WT was loaded at 12.5 μg to reduce underexposure.(C) K→A and Y→A mutants blotted with antibody 1; loading was normalized to 25 μg/sample.(D)Table of antibodies and their binding epitopes with VTG underlined.

Figure S5 .
Figure S5.Quantification of native agarose gels.Representative gel (same CsoS2 agarose gel that is in Figure3B), with boxes drawn around quantified areas.Blue boxes indicate areas of the gel that were quantified to measure shell migration, and yellow boxes indicate areas of the gel that were quantified to measure BSA migration.For shell, density in boxes 2 and 5 were averaged and subtracted as background, since there was no shell present in these samples.For BSA, density in boxes 1, 2, and 3 were averaged and subtracted as background, since no BSA was present in these samples.Three gels for each sample were run and quantified in this way, which is summarized in Figure3C.

Figure S6 .
Figure S6.Shell protein CsoS1A binds CsoS2 with high affinity.Turbidity assay at 10 minutes of the indicated constructs with CsoS1A.Data are the same as in Figure 3C, but with CsoS2B shown for comparison (A) and no shell controls (A and B).Unobservable error bars are smaller than the datapoint icon.All samples contained 9 μM CsoS1A (except for the 0 μM shell sample).Concentrations of MR variants were: 0, 4.5, 9, 12, 14, 16, and 18 μM.The 0 μM shell control had 18 μM of MR.For CsoS2, concentrations tested were 4.5 and 14 μM.The 0 μM shell control had 14 μM of CsoS2.

Figure S7 .
Figure S7.Smaller condensates at 5 minutes after mixing.Fluorescence microscopy of the indicated CsoS2 / MR protein variants with added shell (CsoS1A), imaged at 5 minutes post mixing.All CsoS2 / MR variants are labeled in pink, shell is labeled in green, and the merge appears white at equally overlapping intensities.All proteins are at a final concentration of 10 μM.Scale bar is 20 μm.

Figure S8 .
Figure S8.Larger condensates at 30 minutes after mixing (replicate set).Replicate set of micrographs in Figure 4 of the indicated CsoS2 / MR protein variants with added shell (CsoS1A), imaged at 30 minutes post mixing.All CsoS2 / MR variants are labeled in pink, shell is labeled in green, and the merge appears white at equally overlapping intensities.All proteins are at a final concentration of 10 μM.Scale bar is 20 μm.

Figure S9 .
Figure S9.Over 30 minutes condensate area increases for both CsoS2 and wtMR, but condensates develop into different shapes.(A) Condensate area in μm 2 at 5 minutes and 30 minutes post mixing with shell (CsoS1A) for both CsoS2 and wtMR.(B) Condensate circularity at 5 minutes and 30 minutes post mixing with shell for both CsoS2 and wtMR.Circularity is calculated as 4π*area/perimeter 2 , with 1.0 being a perfect circle and lower values indicating increasing shape elongation.The median is indicated by a black line.Each individual droplet appears as a dot on the plot.Significance of **** is P ≤ 0.0001 in an unpaired t-test.

Figure S10 .
Figure S10.Individual protein controls show no condensate formation.(A) Each indicated protein construct was imaged at 5 minutes and 30 minutes.CsoS2 and MR variants were imaged on the Alexa-647 channel.CsoS1A was imaged on the Alexa-546 channel.All proteins are at a final concentration of 10 μM.All scale bars are 20 μm.(B) Rubisco imaged at 5 minutes on the Alexa-488 channel and phase contrast, and an image merge.Rubisco concentration is 10 μM.Scale bar is 20 μm.

Figure S11 .
Figure S11.Rubisco and CsoS2 form condensates that dissociate over time.CsoS2 is labeled in pink, Rubisco in blue, and the merge is shown in purple.(A) 5 minutes post mixing, two replicate sets, (B) 30 minutes post mixing, two replicate sets.All proteins are at a final concentration of 10 μM.Scale bar is 10 μm.

Figure S13 .
Figure S13.Addition of CsoS1A leads to an increase in droplet size.(A) Average area in μm 2 of CsoS2 + Rubisco condensates and CsoS2 + Rubisco + CsoS1A condensates at 5 minutes and 30 minutes post mixing.The median is indicated by a black line.Each individual droplet appears as a dot on the plot.Significance of **** is P ≤ 0.0001 in an unpaired t-test.(B) Average number of condensates per micrograph, which measured 83.2 x 83.2 μm.The average is written above each bar.Significance of *** is P ≤ 0.001 in an unpaired t-test.Ns, no significance.

Figure S14 .
Figure S14.2% PEG leads to condensate formation in some controls.CsoS2 and MR variants are shown in pink, CsoS1A is shown in green, and Rubisco is shown in blue.CsoS2 and MR variants were imaged on the Alexa-647 channel, CsoS1A on the Alexa-546 channel, and Rubisco on the Alexa-488 channel.Images were taken at 5 minutes and 30 minutes post mixing.All proteins are at a final concentration of 10 μM with 2% PEG-6000.Scale bar is 20 μm.

Figure S15 .
Figure S15.2% PEG leads to some condensate formation of CsoS2 and MR variants with shell at 5 minutes post-mixing.Indicated CsoS2 / MR protein variants with added CsoS1A and 2% PEG-6000, imaged at 5 minutes post mixing.All CsoS2 / MR variants are labeled in pink, CsoS1A is labeled in green, and the merge appears white at equally overlapping intensities.All proteins are at a final concentration of 10 μM.Scale bar is 20 μm.

Figure S16 .
Figure S16.2% PEG leads to some condensate formation of CsoS2 and MR variants with shell at 30 minutes post-mixing.Indicated CsoS2 / MR protein variants with added CsoS1A and 2% PEG-6000, imaged at 30 minutes post mixing.All CsoS2 / MR variants are labeled in pink, CsoS1A is labeled in green, and the merge appears white at equally overlapping intensities.All proteins are at a final concentration of 10 μM.Scale bar is 20 μm.

Figure S17 .
Figure S17.2% PEG leads to CsoS2 and Rubisco condensate formation at 5 and 30 minutes post-mixing.CsoS2 and Rubisco with 2% PEG-6000.CsoS2 is labeled in pink, Rubisco is labeled in blue, and the merge appears purple at equally overlapping intensities.All proteins are at a final concentration of 10 μM.Scale bar is 20 μm.

Figure S19 .
Figure S19.Adding shell to Rubisco + CsoS2 nucleates condensate formation.(A) Gasket setup; the gasket was affixed to a microscope slide, and protein solution was added with a pipette into the pore indicated by the yellow arrow.(B) Experimental schematic.Rubisco and CsoS2 were pre-incubated for 10 minutes before adding shell at time 0. Images were collected over 30 minutes.(C) Still micrographs from the addition experiment.Droplets appear in the focus plane as they condense and adhere to the slide.Larger micrographs show merged channels.Smaller micrographs from select timepoints show each individual channel: CsoS2 in pink, shell in green, and Rubisco in blue.Final concentrations are 8.3 μM Rubisco, 8.3 μM CsoS2, and 17.5 μM shell.Times were rounded to the nearest 30 seconds.Scale bar is 10 μm.

Figure S20 .
Figure S20.Shell + CsoS2 nucleates condensate formation, and Rubisco does not appear to add to condensates.(A) Experimental schematic.Shell and CsoS2 were pre-incubated for 2.5 minutes before adding Rubisco at time 0. Images were collected over 15 minutes.(B) Still micrographs from the addition experiment.Droplets appear in the focus plane as they condense and adhere to the slide.Larger micrographs show merged channels.Smaller micrographs from select timepoints show each individual channel: CsoS2 in pink, shell in green, and Rubisco in blue.Final concentrations are 10 μM Rubisco, 10 μM CsoS2, and 17.5 μM shell.Times were rounded to the nearest 30 seconds.Scale bar is 10 μm.

Figure S21 .
Figure S21.Adding CsoS2 to Rubisco + Shell nucleates condensate formation.(A) Experimental schematic.Shell and Rubisco were pre-incubated for 3.5 minutes before adding Rubisco at time 0. Images were collected over 14 minutes.(B) Still micrographs from the addition experiment.Droplets appear in the focus plane as they condense and adhere to the slide.Larger micrographs show merged channels.Smaller micrographs from select timepoints show each individual channel: CsoS2 in pink, shell in green, and Rubisco in blue.Final concentrations are 10 μM Rubisco, 10 μM CsoS2, and 17.5 μM shell.Times were rounded to the nearest 30 seconds.Scale bar is 10 μm.

Figure S22 .
Figure S22.CTD repeat consensus motif.Comparison between the MR repeat consensus motif and the CTD repeat consensus motif.Sequence logo of the MR repeat was generated from an alignment of 1662 MR repeats identified across 272 dereplicated CsoS2 sequences.Sequence logo of the CTD repeat was generated from an alignment of 528 CTD repeats identified across the same 272 dereplicated CsoS2 sequences.Blue is basic, red is acidic, green is polar/small, black is hydrophobic, yellow is cysteine, purple is aromatic.

Table S1 .
Strains used in H. neapolitanus growth experiments.

Table S2 .
Sequences used in this study.