ROCKET contains two CDL binding sites with different structural implications.

(A) Structure of ROCKET (PDB ID 6B85) in the membrane, with one protein subunit highlighted in purple and phosphate headgroups fo the membrane shown as orange spheres. The structure was obtained from MemProtMD.

(B) Top view of CDL binding to Site 1 taken from a 10 µs snapshot of a CG-MD simulation. The poses was converted to atomistic using CG2AT2 (21) to the CHARMM36m force field (22). CDL is shown as spacefill, and residues R9, K10, W12, and R13 as sticks. The CDL-binding subunit is highlighted in purple.

(C) Design of the ROCKETAAXWA variant. Site 1 on helix 1 of ROCKET is shown on the left (purple), and ROCKETAAXWA with the mutations R9A/K10A/R13A on the right (blue).

(D) CG-MD-derived CDL densities around a heterotetramer composed of two ROCKET subunits (left) and two ROCKETAAXWA subunits (right). Units are number density. Site 1 on ROCKET and Site 2 on ROCKETAAXWA are highlighted by dashed boxes. R9, K10, W12, and R13 are shown as spheres (basic in blue, aromatic in orange). Densities are computed over 5 × 10 µs simulations.

(E) Top view of CDL binding to Site 2 in ROCKETAAXWA following CG-MD and converted to atomistic as per panel B. Interacting residues W12, M16, and R66 on the neighboring subunit (grey) are shown as sticks.

(F) Setup of gas-phase MD simulations for dissociation of ROCKET and ROCKETAAXWA with and without lipids.

(G) Plots of the integral of the force required to separate helix 1 and 2 (d = 1.1 nm), for ROCKET (purple) (p=3.85*10-7) and ROCKETAAXWA (blue) (p=2.93*10−8) with and without bound CDL show a more pronounced increase in stability for CDL-bound ROCKETAAXWA compared to ROCKET (two-tailed t-test with n=20).

(H) Snapshots from gas-phase MD simulations show broad interactions of CDL across the subunits of lipid-bound ROCKETAAXWA (blue) and more localized interactions with fewer intermolecular contacts for ROCKET (purple).

nMS analysis of lipid binding and lipid mediated stabilization of ROCKET and ROCKETAAXWA.

(A) Schematic depiction of electrospray ionization (ESI) process for proteo-liposomes, leading to the ejection of protein-lipid complexes into the gas-phase.

(B) A representative mass spectrum of ROCKET released from proteoliposomes shows tetramers with 1-3 bound CDL molecules, as judged by the characteristic mass shift of 1.4 kDa, as well as additional lipids with molecular weights between 700 and 800 Da.

(C) Release of ROCKETAAXWA from proteoliposomes shows retention of 1-3 CDL molecules per tetramer. The reduced lipid adduct intensity compared to ROCKET indicates reduced lipid binding and/or complex stability.

(D) Schematic illustrating the process of gas-phase subunit unfolding and ejection from ROCKET tetramers at increasing collision energies.

(E) A nMS assay to assess CDL-mediated stabilization of ROCKET and ROCKETAAXWA. Simultaneous dissociation of ROCKET and ROCKETAAXWA leads to the ejection of unfolded monomers, which can be quantified by nMS (top row). Addition of CDL to the same mixture results in lipid binding to tetramers. If CDL binding stabilizes one tetrameric variant more than the other, the amount of ejected monomers will be reduced accordingly (bottom row).

(F) Representative mass spectrum of ROCKET and ROCKETAAXWA at a collision voltage of 220 V. Intact tetramers are seen in the middle, ejected monomers and stripped trimers are seen in the low and high m/z regions, respectively.

(G) Zoom of the low m/z region of a mixture of 25 µM each of ROCKET and ROCKETAAXWA with a collision voltage of 200 V before (left) and after (right) the addition of 50 µM CDL. The three main charge states for both variants can be distinguished based on their mass difference. Addition of CDL reduces the intensity of the ROCKETAAXWA monomer peaks compared to ROCKET (dashed line).

Assessment of lipid-mediated stabilization effects on ROCKET variant.

(A) Principle for pairwise analysis of protein stabilization by CDL. Peaks representing monomers released of protein A and B display intensity changes upon lipid addition. Plotting the peak intensities as a ratio of B to total protein (A + B) shows an increase upon CDL addition, if A is stabilized more than B.

(B) Residues involved in CDL headgroup binding to ROCKET and ROCKETAAXWA were derived from CG-MD simulations (Fig 1) and are shown based on AlphaFold2 models as top view, with the area occupied by CDL as a dashed rectangle, and as side view schematic. A single subunit is colored (purple for ROCKET and blue for ROCKETAAXWA).

(C) Plotting the peak intensity ratios of ROCKET to (ROCKETAAXWA + ROCKET) in the presence and absence of CDL shows a decrease in ROCKETAAXWA monomers when CDL is added (p=0.0007, two-tailed t-tests with n=4).

(D) The CDL binding site and location of the A61P mutation (yellow) mapped on the AlphaFold2 model of ROCKETA61P and shown as a schematic as a side view below. Intensity ratios show significantly more pronounced stabilization of ROCKETAAXWA than ROCKETA61P (p=0.0084, two-tailed t-tests with n=4).

(E) Introduction of a second CDL binding site in the D7A/S8R variant (orange) mapped on the AlphaFold2 model of ROCKETD7A/S8R and shown as a schematic as a side view below. The shift in intensity ratios show that ROCKETAAXWA is still stabilized to a greater extent (p=0.0019, two-tailed t-tests with n=4), but with a smaller margin than ROCKET or ROCKETA61P.

(F) The R66A mutation, designed to disconnect helix 1 from the tetrameric protein core, results in an outward rotation of the CDL binding site, as shown in the AlphaFold2 model (green) and the side view schematic. Intensity ratios show no change upon CDL addition, suggesting that ROCKETR66A is stabilized to a similar extent as ROCKETAAXWA (p=0.8113, two-tailed t-tests with n=4).

(G) Conceptual diagram depicting structural features that promote CDL-mediated stabilization. Distributing the residues that interact with the lipid headgroup, usually basic and aromatic residues, between two helices, as well as Involvement of flexible protein segments, indicated by an outward movement of the right helix, also enhances stabilization by CDL.

Identification of a functional CDL binding site in the GlpG membrane protease.

(A) Distances between basic residues within the same CDL binding site and their occurrence in E. coli membrane proteins. GlpG contains a CDL binding site with two basic residues separated by 75 positions. Site 1 in ROCKET (four positions) is indicated for reference.

(B) Topology diagram of GlpG indicating the locations of basic (R; arginine, L; lysine) and aromatic (W; tryptophan, Y; tyrosine) residues that interact with CDL.

(C) GFP-thermal shift assay of GlpG in detergent or in the presence of PE, PC, SM, PG, or CDL. The average fluorescence intensity (FI) indicates the fraction of soluble protein before (unheated) or after heating to 63 °C and removing precipitated protein. Data are normalized against the non-heated control in detergent. All measurements were performed in triplicates (n = 3).

(D) CG-MD-derived Lipid density plots for CDL (left) and POPG (right) around the transmembrane region of GlpG (PDB ID 2IC8) viewed from the cytoplasmic side. Units are number density. CDL, but not POPG, exhibits preferential binding to the R92-K167 site. The backbones of R92, K67, and Y160 are shown as spheres (basic in blue, aromatic in orange).

(E) nMS spectra of GlpG released from detergent micelles show a 1.4 kDa adduct which can be removed through collisional activation of the protein.

(F) GlpG proteolytic cleavage rates (specific activity, by normalizing to the GFP signal) for soluble (extramembrane) substrate (KSp63) or transmembrane substrate (KSp96, transmembrane helix shaded) in the presence of PG or CDL. Both lipids increase the cleavage rate for the soluble substrate by approx. 40%. For the transmembrane substrate, addition of PG causes a moderate reduction in cleavage activity, whereas CDL causes near-complete inhibition of cleavage.