Non-canonical Shedding of TNFα by SPPL2a Is Determined by the Conformational Flexibility of Its Transmembrane Helix

Abstract

Ectodomain (EC) shedding defines the proteolytic removal of a membrane protein EC and acts as an important molecular switch in signaling and other cellular processes. Using tumor necrosis factor (TNF)α as a model substrate, we identify a non-canonical shedding activity of SPPL2a, an intramembrane cleaving aspartyl protease of the GxGD type. Proline insertions in the TNFα transmembrane (TM) helix strongly increased SPPL2a non-canonical shedding, while leucine mutations decreased this cleavage. Using biophysical and structural analysis, as well as molecular dynamic simulations, we identified a flexible region in the center of the TNFα wildtype TM domain, which plays an important role in the processing of TNFα by SPPL2a. This study combines molecular biology, biochemistry, and biophysics to provide insights into the dynamic architecture of a substrate's TM helix and its impact on non-canonical shedding. Thus, these data will provide the basis to identify further physiological substrates of non-canonical shedding in the future.

Keywords: Biochemistry; Biophysics; Structural Biology.

Graphical abstract

Figure 1

SPPL2a Cleaves Full-Length TNFα Independent of ADAM10/17-Mediated Shedding (A) Schematic representation of TNFα processing. Full-length (FL) TNFα is shedded by ADAM10/17 (canonical shedding). The remaining N-terminal fragment (NTF) is sequentially cleaved by SPPL2a or SPPL2b producing several intracellular domains (ICDs) and a secreted peptide (C-peptide). In addition, SPPL2a cleaves TNFα FL directly before continuing with the processive cleavage (non-canonical shedding). (B) Non-canonical TNFα shedding by endogenous SPPL2a. TNFα ectodomains (ECs) were immunoprecipitated from conditioned media of the HEK293 cell lines expressing endogenous (en.) levels of SPPL2a, SPPL2b, both or none (ko) and visualized utilizing the C-terminal V5-tag. sTNFα(L2), TNFα EC secreted by SPPL2 mediated shedding; sTNFα, TNFα EC secreted by ADAM mediated shedding. (C) Ectopic SPPL2a expression (SPPL2a ex.) significantly increases sTNFα(L2) secretion. TNFα ECs were directly analyzed from conditioned media. sTNFα(L2) was quantified relative to sTNFα from respective Western blots by densitometric analysis and normalized to untreated controls, n = 3. Data are represented as mean ± SD. Statistical significance was calculated applying an unpaired, two-sided Student's t test. ∗p < 0.05. The Western blot shows one representative experiment. (D) sTNFα(L2) secretion is blocked by an SPP/SPPL-specific inhibitor. TNFα ECs from conditioned media of the indicated HEK 293 cell lines were isolated and detected as in (C). Cells were treated with either 50μM Z-LL₂-ketone (+(ZLL)₂) to inhibit SPPL2 catalytic activity or with DMSO as a control (-(ZLL)₂). (E) sTNFα(L2) secretion is not affected by ADAM-specific inhibitors. TNFα ECs from conditioned media of the indicated HEK 293 cell lines were isolated and detected as in (C). Cells were treated with either 5μM GI 254023X and 1μM BMS-561395 (+AI) to inhibit ADAM proteases or with DMSO as a control (-AI). (F) Increased secretion of sTNFα does not affect sTNFα(L2) secretion. TNFα ECs from conditioned media of the indicated HEK 293 cell lines were isolated and detected as in (C). Cells were treated with either 1μM phorbol 12-myristate 13-acetate (PMA), 50μM (ZLL)₂, 5μM GI 254023X, and 1μM 561395 (AI) or DMSO as control. (G) Quantification of (F). sTNFα(L2) was quantified relative to intracellular TNFα FL by densitometric analysis from Western blots as depicted in (F) and normalized to untreated controls; n = 3. Data are represented as mean ± SD. Statistical significance was calculated applying a two-way ANOVA. ∗p < 0.05.

Figure 2

Processing of TNFα by SPPL2a Compared to SPPL2b (A) Amino acid sequences and schematic representation of all TNFα variants used. Arrows indicate the peptides detected by mass spectrometry in (E). (B) C-peptide secretion in SPPL2-expressing cells. C-peptides were immunoprecipitated from conditioned media of HEK293 cell lines expressing exogenous (ex.) levels of SPPL2a, SPPL2b, or none of the proteases (ko) and ectopically expressing the TNFα NTF using the polyclonal V5 antibody. For visualization on Western blot, the monoclonal V5 antibody was employed. (C) ICD generation in SPPL2-expressing cells. Membranes of the indicated HEK293 cell lines ectopically expressing TNFα FL were isolated. Visualization of intracellular TNFα species was done by the monoclonal Flag antibody. Calnexin served as a loading control. (D) Processive turnover of TNFα by SPPL2 proteases. Membranes of HEK293 cell lines co-expressing either SPPL2a or SPPL2b and TNFα FL were incubated for the indicated time periods. ICD conversion was monitored using Western blot and the Flag M2 antibody for detection. Calnexin served as a loading control. (E) N-terminal SPPL2 cleavage sites in TNFα. Mass spectrometric analysis of ICDs generated from SPPL2a or SPPL2b, respectively. Numbers indicate the position of the most C-terminal amino acid of the respective cleavage product. ∗ marks background peak also present in control (see Figure S2). (F) C-terminal SPPL2 cleavage sites in TNFα. Mass spectrometric analysis of C-peptides generated from SPPL2a or SPPL2b, respectively. Numbers indicate the position of the most N-terminal amino acid of the respective cleavage product.

Figure 3

Cleavage Sites of SPPL2a-Mediated Non-Canonical TNFα Shedding (A) Amino acid sequence and schematic representation of TNFα FlagTEV. Arrows indicate the cleavage sites of the respective proteases. (B) Processing of TNFα wt and TNFα FlagTEV is essentially similar. Membranes of HEK 293 cells co-expressing SPPL2a (SPPL2a ex.) and the indicated TNFα variant were analyzed on Western blot using the monoclonal Flag M2 antibody to detect all intracellular TNFα fragments. The respective conditioned media were visualized with the monoclonal V5 antibody to detect secreted TNFα EC variants. (C and D) Cleavage sites of the secreted TNFα ectodomain variants. Secreted EC from HEK 293 cells co-expressing SPPL2a and TNFα FlagTEV were treated with 50μM (ZLL)₂ (D) or DMSO (C). Numbers indicate the position of the most N-terminal amino acid of the respective cleavage product. Peaks from 66 to 77 reflect cleavage of TNFα by ADAM proteases. Peaks from 50 to 55 reflect cleavage of TNFα by SPPL2a. + marks peak of unknown identity

Figure 4

Modulating Non-Canonical Shedding by Mutations in the TM Domain (A) Mutations in the TM domain of TNFα affect non-canonical shedding by SPPL2a. Secreted TNFα from media of cells co-expressing SPPL2a (SPPL2a ex.), and the respective TNFα mutant was analyzed using the V5 antibody. Detection of TNFα FL from membranes served as a transfection control and calnexin as a loading control. (B) Quantification of (A). sTNFα(L2) was quantified relative to sTNFα from respective Western blots by densiometric analysis and normalized to TNFα wt, n = 3. Data are represented as mean ± SD. Statistical significance was calculated applying an unpaired two-sided Student's t-test. ∗p < 0,05. (C) Insertion of a triple L motif reduces SPPL2a shedding. sTNFα from TNFα wt and the AGA/LLL variant was detected with the V5 antibody in media of cells co-expressing SPPL2a. (D) Quantification of (C). sTNFα(L2) was quantified relative to sTNFα from respective Western blots by densiometric analysis and normalized to TNFα wt, n = 3. Data represented as mean ± SD. Statistical significance was calculated applying an unpaired Student's t-test. ∗p = 0,01.

Figure 5

Cleavage Sites of SPPL2a in TNFα Mutants Are Not Changed Secreted C-peptide was immunoprecipitated from media of HEK293 cells co-expressing SPPL2a with anti-Flag M2 affinity gel and analyzed via mass spectrometry. Numbers indicate the position of the most N-terminal amino acid of the respective cleavage product.

Figure 6

Analysis of TNFα TM Helix Flexibility by DHX (A) Calculated Gibb's free energy differences ΔG of amide H-bond disruption of the TNFα wt TM helix. Error bars correspond to standard confidence intervals (calculated from the errors of fit in k_exp determination). Broken lines denote the major initial SPPL2a cleavage sites. (B) Differences ΔΔG between TNFα mutant and wt amides (±confidence intervals of mutants). Confidence intervals of wt amides are plotted in light gray on the x axis. In the case of the colored bars, the confidence intervals of mutant and wt are not overlapping while they do overlap in the case of the gray bars.

Figure 7

3D Structural Analysis of the TNFα TM Segment by NMR (A) Solution structures of TNFα wt (black), TNFα S34P (green), and TNFα AGALLL (orange). The A42/G43/A44 motif is represented in red (wt and S34P), respectively, in pink in AGA/LLL. P34 in S34P is shown in pink. (B) Structural bundles of TNFα wt (black), TNFα S34P (green), and TNFα AGA/LLL (orange). Top and side views of the 20 best structures superimposed onto the backbone of C-terminal helix from residue 42 to 53. The A42/G43/A44 motif is represented in red. Amino acid exchanges in the respective mutants are highlighted in magenta. (C) Backbone conformations of NMR structures. Top section: Probability that a residue is part of a single hinge (type Bend 1) or a pair of hinges (type Bend 2). Dashed lines indicate cleavage sites. Bottom section: Bend 1 and Bend 2 subsets aligned to an ideal reference helix (gray cylinder). The hinges (green) are coordinating bending between helical segments. Bend 1 structures were aligned to the helical C-terminal segment of the reference structure (gray cylinder); Bend 2 conformations were aligned to the helical segment between the two hinges. The location of S34 and S37 Cα atoms is highlighted as purple spheres. Mutation sites are drawn in space fill mode (Leu: red, Pro: black). Exemplary conformations illustrating the fundamental motions are explained in more detail in (Götz et al., 2019b; Götz and Scharnagl, 2018).

Figure 8

TNFα TM Helix Dynamics in a POPC Bilayer from MD Simulations (A) Residue insertion depths relative to the lipid phosphate heads. Gray areas represent the location of the phosphate heads. The dots indicate the average Cα position (larger dots specify helical residues as determined by secondary structure analysis, see Figure S7). Dashed lines indicate cleavage sites. (B) Distribution of tilt angle between the TM-helix (F36-C49) and the membrane normal. The colored areas show the 95% CI. (C) Representative structures illustrating the orientation in the membrane. Red spheres show lipid phosphates. (D) Backbone conformations of TNFα wt and mutant TM helices characterized by the probability that a residue is part of a single hinge (types Bend 1, Twist 1) or a pair of hinges (types Bend 2, Twist 2). Dashed lines indicate cleavage sites.