Alpha-1-B glycoprotein (A1BG) inhibits sterol-binding and export by CRISP2

Abstract

Proteins belonging to the CAP superfamily are present in all kingdoms of life and have been implicated in various processes, including sperm maturation and cancer progression. They are mostly secreted glycoproteins and share a unique conserved CAP domain. The precise mode of action of these proteins, however, has remained elusive. Saccharomyces cerevisiae expresses three members of this protein family, which bind sterols in vitro and promote sterol secretion from cells. This sterol-binding and export function of yeast Pry proteins is conserved in the mammalian cysteine-rich secretory protein (CRISP) proteins and other CAP superfamily members. CRISP3 is an abundant protein of the human seminal plasma and interacts with alpha-1-B glycoprotein (A1BG), a human plasma glycoprotein that is upregulated in different types of cancers. Here, we examined whether the interaction between CRISP proteins and A1BG affects the sterol-binding function of CAP family members. Coexpression of A1BG with CAP proteins abolished their sterol export function in yeast and their interaction inhibits sterol-binding in vitro. We map the interaction between A1BG and CRISP2 to the third of five repeated immunoglobulin-like domains within A1BG. Interestingly, the interaction between A1BG and CRISP2 requires magnesium, suggesting that coordination of Mg²⁺ by the highly conserved tetrad residues within the CAP domain is essential for a stable interaction between the two proteins. The observation that A1BG modulates the sterol-binding function of CRISP2 has potential implications for the role of A1BG and related immunoglobulin-like domain containing proteins in cancer progression and the toxicity of reptile venoms containing CRISP proteins.

Keywords: CAP protein superfamily; Saccharomyces cerevisiae, snake venom toxins; cysteine-rich secretory protein (CRISP); immunoglobulin domain (Ig); oprin; protein-protein interactions; sterol export.

Figure 1

A1BG inhibits sterol export by CAP family proteins. A, schematic representation of the experimental system. Heme-deficiency of Saccharomyces cerevisiae mimics anaerobic conditions and allows for labeling of cells with exogenously supplied [¹⁴C]-cholesterol. Deletion of the sterol deacetylase Say1 results in the accumulation of acetylated [¹⁴C]-cholesterol, which is then secreted out of the cells by binding to the yeast CAP proteins Pry1 and Pry2. The block of secretion of [¹⁴C]-cholesterol acetate of a quadruple mutant (4Δ; hem1Δ say1Δ pray1Δ pry2Δ) is complemented by a plasmid-borne copy of a mammalian CAP family proteins such as CRISP. Coexpression of a protein which interacts with CRISP, such as A1BG is being used to assess whether this protein-protein interaction (CRISP-A1BG) affects the sterol binding and export function of CRISP. B, export of acetylated cholesterol is blocked in cells expressing A1BG. Acetylation and export of [¹⁴C]-cholesterol was examined in hem1Δ say1Δ double mutant cells and in quadruple mutant cells lacking the endogenous CAP family members Pry1 and Pry2 (hem1Δ say1Δ pry1Δ pry2Δ). Strains expressing the indicated CAP proteins (PRY1, CRISP3, or CRISP2) from a plasmid (pRS416) and coexpressing A1BG or bearing an empty control plasmid (pRS414) were cultivated in the presence of [¹⁴C]-cholesterol. Lipids were extracted from cell pellet (P) and culture supernatant (S), separated by TLC, and visualized by phosphorimaging. The position of free cholesterol (FC), cholesterol acetate (CA), steryl esters (STE), and an unidentified lipid (∗) are indicated to the right. C, quantification of cholesterol acetate export. The export index represents the relative levels of cholesteryl acetate exported by cells. Export inhibition by A1BG was consistent across three independent experiments, indicating a robust and reproducible interaction between A1BG and CAP family proteins. Data correspond to means ± S.D. of 3 independent experiments and statistical significance is indicated: ∗∗p ≤ 0.01; ∗∗∗∗p ≤ 0.0001 (one-way ANOVA with Tukey’s post hoc test). Precise p-values of all statistical analyses are given in Tables S1. D, expression of A1BG does not block the synthesis or secretion of CAP family members. Proteins were TCA precipitated from the cell pellet (P) and the culture medium (S) of cells expressing HA-tagged Pry1, CRISP2, or CRISP3 in the presence (+) or absence (−) of FLAG-tagged A1BG and analyzed by Western blotting. Pry1 is detected as a high molecular weight glycosylated protein in the culture supernatant. A1BG fused to the N-terminal signal sequence of pre-pro alpha factor is present as a pre-pro form in the culture supernatant. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; TCA, trichloroacetic acid.

Figure 2

A1BG binds CRISP2 in vitro and blocks sterol binding. A, the direct interaction between A1BG and CRISP2 was assessed by MST. Purified and fluorescently labeled CRISP2 protein was incubated with increasing concentrations of unlabeled A1BG and complex formation was analyzed. MST measurements were performed in triplicates, and the corresponding dissociation constant (K_d) is indicated. B–D, sterol-binding by CRISP2 in the absence (panel B) or the presence of A1BG (panel D) was assessed by MST. CRISP2 binds cholesterol sulfate in the low micromolar range, but this binding is blocked in the presence of A1BG, which by itself does not bind cholesterol sulfate (panel C). Measurements were performed in triplicates and the corresponding dissociation constants (K_d) are indicated. N/A; not applicable. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; MST, microscale thermophoresis.

Figure 3

In silico docking of CRISP2 to A1BG.A, schematic representation of CRISP2 and A1BG. The CAP domain (green box) and the ion channel regulatory domain (ICR, pink box) of CRISP2 are depicted. The five immunoglobulin (Ig) domains of A1BG are color coded. Amino acid positions of domain boundaries are indicated. The structures of both proteins as predicted by AlphaFold2 are shown below in a combined ribbon/surface representation. The large CAP cavity within the CAP domain of CRISP2 is indicated as is the ring-shaped alignment of the five Ig domains of A1BG. B, the predicted interaction of the CAP domain of CRISP2 (green) with A1BG is shown using a combined ribbon/surface representation. C, the area of surface-to-surface interaction between the CAP domain of CRISP2 and the individual Ig domains of A1BG are indicated in light colors. The largest surface interaction occurs between the Ig3 domain of A1BG and the CAP domain of CRISP2, followed by that of Ig1 and Ig5. D, table depicting the predicted interaction between CRISP2 and individual Ig domains of A1BG. Free energy (ΔG) and affinities (K_d) of interactions between CRISP2 and individual Ig domains of A1BG as predicted by PRODIGY are listed. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; PRODIGY, PROtein binDIng enerGY prediction.

Figure 4

Expression of the Ig3 domain of A1BG inhibits cholesterol export by Pry1 or CRISP2. A, sterol export by yeast Pry1 and mammalian CRISP2 in the presence of the five different Ig domains, Ig1-Ig5, of A1BG. Quadruple mutant cells (hem1Δ say1Δ pry1Δ pry2Δ) expressing different combination of either empty plasmids (pRS414, pRS416), plasmids containing the indicated Ig domains of A1BG, or plasmids for the expression of the CAP proteins Pry1 and CRISP2 were labeled with [¹⁴C]-cholesterol. Lipids were extracted from both the cell pellet (P) and the culture supernatant (S), separated by TLC, and visualized by phosphorimaging. The position of free cholesterol (FC), cholesterol acetate (CA), steryl esters (STE), and an unidentified lipid (∗) are indicated to the right. B, export of radiolabeled cholesterol acetate is plotted as export index in the graph. Data correspond to means ± S.D. of three independent determinations and statistical significance is indicated: ∗∗∗p ≤ 0.001; ∗∗∗∗p ≤ 0.0001 (one-way ANOVA with Tukey’s post hoc test). C, expression of the Ig domains of A1BG does not block the synthesis or secretion of the CAP family members Pry1 or CRISP2. Proteins were TCA precipitated from the cell pellet (P) or the culture supernatant (S) of cells expressing HA-tagged Pry1 or CRISP2, in the absence (pRS414, empty plasmid) or the presence of FLAG-tagged Ig domains of A1BG and analyzed by Western blotting. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; HA, hemagglutinin; Ig, immunoglobulin; TCA, trichloroacetic acid.

Figure 5

The Ig3 domain of A1BG displays high affinity interaction with CRISP2 in vitro and in vivo. A, the in vitro interaction between three different Ig domains of A1BG, Ig1, Ig3, and Ig4 and CRISP2 was assessed by microscale thermophoresis. Purified and fluorescently labeled CRISP2 protein was incubated with increasing concentrations of unlabeled purified Ig domains of A1BG, Ig1, Ig3, and Ig4, and complex formation was analyzed. Measurements were performed in triplicates and the corresponding dissociation constant (K_d) is indicated. N/A; not applicable. B, the in vivo interaction between three different Ig domains of A1BG, Ig1, Ig3, and Ig4 and CRISP2 was assessed by co-immunoprecipitation. Cells expressing FLAG-tagged Ig domains of A1BG, Ig1, Ig3, or Ig4 (a) and cells coexpressing FLAG-tagged Ig domains together with HA-tagged CRISP2 (b) were cultivated, and the interaction between the Ig domains of A1BG and CRISP2 in the culture medium was analyzed by immunoprecipitation with an anti-HA antibody followed by Western blotting to detect both HA-tagged CRISP2 and FLAG-tagged Ig domains. C, quantification of the interaction between the Ig domains of A1BG and CRISP2. The interaction between Ig1, Ig3, or Ig4 and CRISP2 detected by Co-IP were quantified and plotted as ratio between the signal obtained from cells expressing both proteins, CRISP2-HA and Ig-FLAG (+), divided by the background signal from cells lacking CRISP2-HA (−). Data represent mean ± S.D. of 5 determinations and statistical significance is indicated: ∗∗p ≤ 0.01; ∗∗∗∗p ≤ 0.0001 (one-way ANOVA with Tukey’s post hoc test). A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; HA, hemagglutinin.

Figure 6

The Ig3 domain of A1BG blocks sterol binding by CRISP2 in vitro. A–D, sterol-binding by CRISP2 in the absence (panel A) or the presence of purified Ig domains of A1BG (panels B–D) was assessed by microscale thermophoresis. CRISP2 bound cholesterol sulfate in the low micromolar range (panel A). Sterol binding by CRISP2 is not affected in the presence of the Ig1 (panel B) or Ig4 (panel D) domains of A1BG but is blocked in the presence of Ig3 (panel C). E–G, the Ig domains by themselves do not bind cholesterol. Measurements were performed in triplicates and the corresponding dissociation constants (K_d) are indicated. N/A; not applicable. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; Ig, immunoglobulin.

Figure 7

The interaction between the Ig3 domain of A1BG and CRISP2 is affected by mutations in the β2-strand of CRISP2. A, Model of the interaction between the Ig3 domain of A1BG and CRISP2. The interaction between the Ig3 domain (blue) and CRISP2 (green) as predicted by in silico docking using AlphaFold. The complex between Ig3 and CRISP2 is mainly stabilized by the interaction between a β-sheet of Ig3 (highlighted in light blue) and the antiparallel β2-strand of CRISP2 (pink) involving the stabilizing residues L98 and Y99. Mutations of both L98 and Y98 to glycine and proline are predicted to disturb the β2-strand of CRISP2 (right-hand panel) resulting in a decreased affinity between Ig3 and CRISP2 as indicated by the dissociation constant predicted by PRODIGY (pred K_d). B and C, mutations within the β2-strand residues L98 and Y99 of CRISP2 affect its interaction with A1BG and Ig3. The binding of the CRISP2^{L98G Y99P} double mutant version with A1BG, the Ig3 domain, and cholesterol sulfate was assessed by MST. The presence of Ig3 still inhibited sterol binding by CRISP2^{L98G Y99P}. Measurements were performed in triplicates and the corresponding dissociation constants (K_d) are indicated. N/A; not applicable. D, export of cholesterol by the CRISP2^{L98G Y99P} mutant version is impaired in the presence of Ig3. Quadruple mutant cells (hem1Δ say1Δ pry1Δ pry2Δ) expressing CRISP2^{L98G Y99P} and carrying either an empty plasmid (pRS414) or a plasmid containing the Ig3 domains of A1BG were labeled with [¹⁴C]-cholesterol. Lipids were extracted from both the cell pellet (P) and the culture supernatant (S), separated by TLC, and visualized by phosphorimaging. The position of free cholesterol (FC), cholesterol acetate (CA), steryl esters (STE), and an unidentified lipid (∗) are indicated to the right. E, quantification of export of radiolabeled cholesterol acetate is plotted as export index. Data correspond to means ± S.D. of three independent determinations and statistical significance is indicated: ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001 (one-way ANOVA with Tukey’s post hoc test). F, expression of the Ig3 domains of A1BG does not affect the synthesis or secretion of the CRISP2^{L98G Y99P} mutant version in yeast. Proteins were TCA precipitated from either the cell pellet (P) or the culture supernatant (S) of cells expressing HA-tagged CRISP2^{L98G Y99P}, in the absence (pRS414, empty plasmid) or the presence of a FLAG-tagged Ig3 domain and analyzed by Western blotting. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; Ig, immunoglobulin; HA, hemagglutinin; L98, leucine residue at position 98; MST, microscale thermophoresis; PRODIGY, PROtein binDIng enerGY prediction; TCA, trichloroacetic acid.

Figure 8

CRISP2 requires magnesium for binding of Ig3, A1BG, or cholesterol.A, model of the CAP domain of CRISP2 showing the highly conserved tetrad residues proposed to coordinate binding of divalent cations. Mg²⁺ is indicated in magenta, the conserved tetrad resides (H82, E96, E117, and H136) are indicated. B and C, binding of the Ig3 (panel B) domain of A1BG or that of full-length A1BG (panel C) by CRISP2 is dependent on magnesium. Binding of Ig3 or A1BG by CRISP2 was assessed by MST in the presence of EDTA (5 mM), EDTA and MgCl₂ (10 mM), or EDTA and ZnCl₂ (20 mM). D, binding of cholesterol sulfate by CRISP2 requires magnesium. Binding of cholesterol sulfate by CRISP2 was assessed by MST in the presence of EDTA, MgCl₂ or ZnCl₂. All MST measurements were performed in triplicates and the corresponding dissociation constants (K_d) are indicated. N/A; not applicable. A1BG, alpha-1-B glycoprotein; CRISP, cysteine-rich secretory protein; Ig, immunoglobulin; MST, microscale thermophoresis.