Split NeissLock with Spy-Acceleration Arms Mammalian Proteins for Anhydride-Mediated Cell Ligation

Abstract

Reactive functional groups may be incorporated into proteins or may emerge from natural amino acids in exceptional architectures. Anhydride formation is triggered by calcium in the self-processing module (SPM) of Neisseria meningitidis FrpC, which we previously engineered for "NeissLock" ligation to an unmodified target protein. Here, we explored bacterial diversity, discovering a related module with ultrafast anhydride formation. We dissected this swift SPM to generate a split NeissLock system, providing a second layer of control of anhydride generation: first mixing N- and C-terminal NeissLock moieties and second adding millimolar amounts of calcium. Split NeissLock generated a minimal fusion tag, permitting binder expression in mammalian cells with complex post-translational modifications and avoiding self-cleavage while transiting the calcium-rich secretory pathway. Employing spontaneous amidation between SpyTag003 and SpyCatcher003, we dramatically accelerated split NeissLock reconstitution, allowing a rapid high-yield reaction to naturally occurring targets. We established a specific covalent reaction to endogenous Epidermal Growth Factor Receptor using split NeissLock via Transforming Growth Factor-α secreted from mammalian cells. Modular ligation was demonstrated on living cells through site-specific coupling of the clot-busting enzyme tissue plasminogen activator or a computationally designed cytokine. Split NeissLock provides a modular architecture to generate highly reactive functionality, with inducibility and simple genetic encoding for enhanced cellular modification.

1

Identifying an ultrafast self-processing module (SPM). (A) SPM undergoes autoproteolysis at Asp-Pro, generating an anhydride. POI is the protein of interest. (B) Schematic of NeissLock. A binding protein genetically fused to SPM docks with a target protein. Upon adding calcium, an anhydride (marked by the red star) is generated on the binding protein, releasing SPM, and enabling covalent coupling to a nucleophile (e.g., lysine, K) on the target. The red line represents an isopeptide bond. (C) Reactivity of SPM homologues. Incubation of different versions of 5 μM OAZ-SPM with 5 μM ODC for 0 or 16 h was done with 10 mM calcium at 37 °C, before SDS–PAGE/Coomassie analysis. A colon indicates covalent coupling. (D) Time course for SPM cleavage. 5 μM OAZ-SPM was mixed with 5 μM ODC for varying times with 10 mM calcium at 37 °C, before SDS–PAGE/Coomassie. (E) SPM cleavage rate for FrpA and FrpC with 10 μM of each partner, after adding 1 mM calcium for the indicated time at 25 °C. (F) Coupling rate was tested as in (E). Plots show mean ± 1 s.d., n = 3.

2

Engineering split NeissLock coupling. (A) Schematic of the split NeissLock. A binding protein genetically fused to the N-terminal fragment of SPM reconstitutes with SPM’s C-terminal fragment, before binding the target protein. Calcium activates anhydride generation, promoting ligation to the target. (B) AlphaFold 3 model of FrpA SPM, color-coded for regions for initial splitting into N-terminal (mauve, residues 300 to 315) and C-terminal (orange, residues 316 to 543) fragments. The reactive aspartate (D*) is shown in a stick format. Ca²⁺ ions are shown as gray spheres. (C) Split NeissLock allows covalent ligation. OAZ-SPM_N315 and SPM_C316 each at 10 μM were incubated ± 10 μM ODC ± calcium at 37 °C for 16 h, before SDS–PAGE/Coomassie. (D) Schematic of the different tested SPM_N and SPM_C fragments. (E) Time course for ligation using SPM_N and SPM_C fragments. OAZ-SPM_N was premixed with SPM_C, before incubating with ODC (each protein at 5 μM) along with calcium for the indicated times at 37 °C. Reaction was analyzed by SDS-PAGE/Coomassie (mean ± 1 s.d., n = 3).

3

Spy-directed split NeissLock. (A) Schematic of Spy-accelerated split NeissLock. A binding protein fused to SPM’s N-terminal fragment and SpyTag003 reacts with SPM’s C-terminal fragment fused to SpyCatcher003, to promote SPM reconstitution before calcium activation. (B) Electrospray-ionization MS of reconstitution and SPM cleavage. OAZ-SPM_N-SpyTag003 was incubated with SPM_C-SpyCatcher003 and analyzed ± calcium. (C) Spy-acceleration of split SPM cleavage. 2 μM OAZ-SPM_N was incubated with 2 μM SPM_C ± SpyTag003/SpyCatcher003 fusion at 37 °C, before adding calcium for the indicated time and SDS-PAGE/Coomassie. (D) Quantification of Spy-accelerated cleavage, based on (C) (mean ± 1 s.d., n = 3). (E) Optimization of the Split Site for Spy-acceleration. Percentage cleavage upon mixing the OAZ-SPM_N-SpyTag003 and SPM_C-SpyCatcher003 variants was displayed as a heat map. 2 μM of each fragment was preincubated for 1 h at 37 °C, before calcium for 5 min (mean of n = 3).

4

Spy-directed split NeissLock for coupling of model therapeutic domains to EGFR. (A) Covalent ligation of tPA to EGFR’s extracellular domain. 3 μM tPA-TGFα-SPM_N-SpyTag003 was reconstituted with 3 μM SPM_C-SpyCatcher003 for 30 min at 25 °C. 1.4 μM amount of sEGFR (the soluble extracellular region of EGFR) was added for 15 min at 37 °C, followed by Ca²⁺ for 1 h at 37 °C. Coupling was analyzed by SDS-PAGE/Coomassie, after PNGase F deglycosylation to simplify banding patterns. (B) Specific coupling of tPA or cytokine domains to EGFR in living cells. A431 cells were labeled for 10 min with 2 mM calcium and 1 μM tPA or Neo2/15 linked to TGFα for split NeissLock. Covalent products were detected by Western blot with anti-TGFα. Controls have R42A TGFα to block EGFR binding, DA-mutated SPM_N to block anhydride formation, or hydroxylamine to inactivate the anhydride. Blotting to GAPDH was the sample processing control.