Differential Membrane Interactions of Saposins A and C: Implications for the Functional Specificity∗ Submitted by:

Saposins are small, heat-stable glycoprotein activators of lysosomal glycosphingolipid hydrolases that derive from a single precursor, prosaposin, by proteolytic cleavage. Three of these saposins (B, C, and D) share common structural features including a lack of tryptophan, a single glycosylation sequence, the presence of three conserved disulfide bonds, and a common multiamphipathic helical bundle motif. Saposin A contains an additional glycosylation site and a single tryptophan. The oligosaccharides on saposins are not required for in vitro activation functions. Saposins A and C were produced in Escherichia coli to contain single tryptophans at various locations to serve as intrinsic fluorescence reporters, i.e. as topological probes, for interaction with phospholipid membranes. Maximum emission shifts, aqueous and solid quenching, and resonance energy transfer were quantified by fluorescence spectroscopy. Amphipathic helices at the amino- and carboxyl termini of saposins A and C were shown to insert into the lipid bilayer to about five carbon bond lengths. In comparison, the middle region of saposins A or C were either embedded in the bilayer or solvent-exposed, respectively. Conformational changes of saposin C induced by phosphatidylserine interaction suggested the reorientation of functional helical domains. Differential interaction models are proposed for the membrane-bound saposins A and C. By site-directed mutagenesis of saposin A and C, their membrane topological structures were correlated with their activation effects on acid beta-glucosidase. These findings show that proper orientation of the middle segment of saposin C to the outside of the membrane surface is critical for its specific and multivalent interaction with acid beta-glucosidase. Such membrane interactions and orientations of the saposins determine the proximity of their activation and/or binding sites to lysosomal hydrolases or lipoid substrates.


INTRODUCTION
Saposins, a family of small (~ 80 amino acids) heat stable glycoproteins, are essential for the in vivo hydrolytic activity of several lysosomal enzymes in the catabolic pathway of glycosphingolipids (1)(2)(3). Four members of saposins, A, B, C, and D, are proteolytically derived from a single precursor protein, termed prosaposin (4)(5)(6)(7)(8). The primary sequences of saposins are highly homologous with ~60% amino acid similarity. In addition, each of the saposins has six conserved cysteines that form three intradomain disulfide bridges whose placements are identical (9). In a native state, five consensus N-glycosylation sequences in prosaposin are occupied by oligosachharide chains (10)(11)(12). Each of saposins B, C, and D has one such sequence, while saposin A has two. Non-glycosylated saposins retain their respective activation effects using in vitro assays (12)(13)(14)(15).
A multiple α-helical bundle motif, characterized by a three conserved disulfide structure and several amphipathic peptides, is found in the saposins, and also in saposin-like proteins and domains, i.e., NK-lysin, surfactant-associated protein B (SP-B), acid sphingomyelinase (ASM), acyloxylacyl hydrolase (AOAH), plant aspartic proteases and pore-forming peptides (amoebapores), [reviewed in (16)]. This common backbone structure was characterized by the Saposins Membrane Interaction -5 -Despite the shared saposin fold structure in solution, saposins and saposin-like proteins have diverse in vivo biological functions. Since all these proteins bind to or interact with lipid membranes, one can speculate that the specific biological functions of saposins and saposin-like proteins are the results of the differential interactions with the biological membrane environments. For instance, SP-B and NK-lysin are saposin-like proteins with completely different biological functions. SP-B plays a crucial role in the rapid adsorption of lipids at the air/water interface (19). NK-lysin is a tumor-lysin and antibacterial polypeptide (20).  (21). The insertion of α-helical peptides and the electrostatic interaction between charged residues of the protein and polar head groups of lipid stabilize SP-B at lipid membrane surface. Lipid-bound NK-lysin model has half of the molecule embedded into the membrane and the more hydrophilic negatively-charged half remains solvent-exposed (22). Most of the positively-charged residues located in an equatorial belt around the saposin fold of the molecule interact with the negatively charged head groups of lipid. Apparently, the different interaction modes of SP-B and NK-lysin with lipid membrane are due to the different charge arrangements and amphipathic properties of helices in these two molecules.
Prosaposin is a multifunctional precursor preprotein. A complete deficiency of prosaposin with a mutation in the initiation codon caused the storage of multiple glycosphingolipid substrates by guest on  http://www.jbc.org/ Downloaded from resembling a combined lysosomal hydrolase deficiency (23). Patients lacking the individual saposins B and C showed a variant form of metachromatic leukodystrophy (24)(25)(26) and Gaucher disease (27,28), respectively. These define the primary physiological functions of saposin B and C in lysosomes. The structural characteristics of these saposins are of great importance to their diverse mechanisms of activation on their respective cognate glycosphingolipid hydrolases.
Mechanistic and kinetic studies of lysosomal enzyme activation by saposins B and C have shown different modes of action. In general, activation by saposin B is via solubilizing and presenting glycosphingolipid substrates to several lysosomal enzymes (2). Saposin C promotes acid βglucosidase activity by inducing in the enzyme conformational change at acidic pH (29)(30)(31). The enzyme activity is thought to be optimized by saposin C via membrane perturbation (32).
However, the binding of saposin C to phospholipid bilayers is poorly characterized. In vitro and ex vivo saposins A and D functions are to enhance the degradation of galactosylceramide and ceramide/sphingomyelin, respectively (12,(33)(34)(35). To date, the precise physiological functions of saposins A and D remain unknown.
In vitro saposin A enhances acid β-glucosidase activity at µM concentration, but isolated saposin C deficiency leads to glucosylceramide storage and a "Gaucher disease-like" phenotype (36,37).
In a variety of cells, the physiological concentrations of saposins are estimated to be in the nmolar range (39) and saposin C provides excellence in vitro activation levels at # 200 nM.
This activation was mediated via a highly specific induced conformational change and multivalent interactions. Importantly, negatively-charged phospholipid bilayers are required for the interactions of saposin A and/or C with acid β-glucosidase in vitro. Consequently, their mechanisms of interaction may be influenced by their differential interaction.
In this study, the differential interactions of saposins A and C with lipid bilayers were evaluated by intrinsic Trp fluorescence spectroscopy. Mutant saposins provided insight into the molecular basis for the conformational requirement of membrane-bound saposins to the specific activation functions. Saposin Preparation  The amino acid sequences of wild-type saposins A and C are in Figure   1. The mature NH 2 -terminal amino acid of the natural human saposins will be designated as 1 and used to reference the position of the amino acids in the fragments. The cDNAs for mutant proteins were generated using the QuickChange site-directed mutagenesis kit, and their sequences were verified by complete DNA sequencing.
Saposins Membrane Interaction -9 - The major protein peak was collected and lyophilized. The protein concentrations were determined as previously described (15).
Vesicle Preparation  Unilamellar vesicles (SUV) were prepared by bath sonication (15,31). Since saposins cause fusion of the vesicles within 2 to 3 minutes of introduction, the selected saposins were incubated for about 20-30 minutes before collection of any fluorescence measurements. The vesicles had ~200 nm diameters (N4 + particle sizer). Using 31 P-NMR spectroscopy in the presence of Mn 2+ quenching (40), lipid dispersions of egg PC in sodium acetate (pH 4.7) with vortex mixing, showed 16-28% 31 P quenching by addition of MnCl 2 (5 mM). Since the 31 P in PC would be distributed to inaccessible sites within multilamellar liposomes (inner leaflet and internal concentric bilayers), the quenching percentage would be expected to be significantly less than 50%. In comparison, egg PC liposomes prepared by the above sonication procedure led to a ~ 50% reduction in 31 P signal upon addition of Mn 2+ . These results are consistent with those expected for unilamellar vesicles.
Saposin Activation Assays  The saposin activities toward acid β-glucosidase were determined fluorometrically (15,31). Assays were conducted in a detergent-free system with  were obtained in the presence of brain phosphatidylserine (BPS) liposomes. Maximal emission wavelength shifts to the blue direction were noted for the Trp substituted saposins compared to those in a lipid-free background (Table I). This parameter for saposin C (S37W) did not change.
Fluorescence intensities also were increased significantly upon Trp-saposin addition to BPS or synthetic PS (18:1,1) liposomes at acidic pH (Fig. 2). These blue-shifts and intensity elevations suggested interaction of saposins with liposomal membranes during protein-lipid complexes formation. None of the Trp-saposin As or Cs showed blue-shifts with the neutral egg phosphatidylcholine (EPC) or saturated phosphatidylserine (PS) liposomes. These results also show that the electrostatic interactions of the saposins with the charged head groups were insufficient to account for the respective spectral shifts. In the absence of phospholipids, saposin A (37W) showed the same maximal emission wavelength (λ EM = 351 nm) as saposin C (S37W).
This indicated that the polarity of middle regions in saposins A and C are similar in lipid-free solution.
Fluorescence quenching experiments of Trp-saposins by aqueous (polar) or hydrophobic solid (nonpolar) quenchers were conducted to determine whether saposins interacted with BPS liposomal membrane at the surface or by insertion. Aqueous fluorescence quenchers, acrylamide, were used to evaluate the accessibility of the tryptophanyl residue in saposins to the aqueous phase in the presence of BPS liposome (Fig. 3A). With the liposomes composed of BPS, the blue-shifts of the Trp-saposin fluorescence were protected effectively from quenching by acrylamide. However, saposin C (S37W), that did not have spectral shift, was largely quenched. In the absence of liposomes, all Trp-saposins were quenched to the same degree and the degree of quenching was much greater than in the presence of BPS. Together with spectral shifting experiments, the results indicated that the middle region of saposin C around 37S was exposed to the aqueous phase while the amino-and carboxyl-terminals regions were associated with membrane lipids. In comparison to saposin C (S37W), wild-type saposin A (37W) was quenched much less by acrylamide in the presence of BPS vesicles (Fig. 3A) (Table I) Saposin C (S37W) was added into mixed BPS/DPS liposomes for FRET analysis (Fig. 4). The emission spectrum of Trp in saposin C and dansyl-groups with BPS/DPS vesicles showed λ EM at 351 and 503 nm (Fig. 4, lines A and B), respectively. FRET of saposin C (S37W) associated with the BPS/DPS membrane is shown in Figure 4  Effects of Phospholipids on Saposin Conformation  Saposin A and C had different topological associations with BPS liposomal membrane. The concordant conformational changes also were studied by CD spectroscopy (Table II). The CD spectral changes in saposin C were obtained with PS (18:1,1) at pH 4.7. A relative increase in the α-helix content and a relative decrease the β-strands were observed (Table II, (Table I), respectively, in the presence of BPS. This result indicates that the Q48N and Q48A/E49A substitutions alter the interactions of saposin C with membranes and that these alterations are associated with a loss of activation function. In the absence of BPS, the λ EM of saposin C (S37W) was 6-13 nm higher than for the additionally mutated saposins (Table I). These spectral shifts of the multiply mutated saposin Cs indicate that the microenvironments surrounding the Trp had different polarities. Interestingly, CD spectra were similar to the native sequence in the absence of phospholipids (Table II). The BPS-induced relative secondary structural changes were greater with saposin C (Q48N) and C (Q48A/E49A) than with wild-type saposin C. These CD spectra changes paralleled the blue-shifts observed with the respective mutated saposin Cs in the presence of various BPS.
Our previous enzyme activation studies highlighted the importance of the COOH region and residues 47 to 60 of saposin C for (44). The residues 47 to 60 in saposin A are highly similar to the corresponding residues in saposin C (Fig. 1B, heavy line). Importantly, saposin A competes with saposin C for the binding sites on the enzyme at nmolar concentrations (15). However, nmolar concentrations of saposin A have no activation effects on the enzyme. The differential membrane interaction of saposin A and C indicated that the orientation of the corresponding domains in membrane-bound saposin A was not same as those in saposin C. Thus, the produced, after a lag phase, about 70-90% of the acid β-glucosidase activation as the wild-type saposin C (Fig. 6). By fluorescence spectroscopy, the increase in enzymatic activation by the mutated saposin As correlated with significant red-shifts in emission spectra in the presence of BPS (Table I). These red-shifts (from W37) suggest that the middle regions of mutant saposin As had become exposed to more polar environments upon binding of BPS membranes. Thus, the membrane topology changes from that of normal saposin A to one that was more similar to that of saposin C and leading to development of the enzyme activation property.

DISCUSSION
The present experiments demonstrate the differential membrane interactions of saposins A and C. Native saposin C was engineered to contain specifically placed Trps as intrinsic fluorescence probes in the amino-or carboxyl-termini and middle regions. These substitutions did not alter the activation properties of these saposin Cs toward acid β-glucosidase. By fluorescence emission and quenching analyses, the saposin C amphipathic helices 1 and 5 inserted, at acidic pH, into outer-leaflet of particular negatively charged phospholipid membranes to a depth of about 5 carbon bonding lengths. A two step binding model is suggested in which saposin C has an initial interaction mediated by electrostatic effects and a second hydrophobic-interaction step that leads to insertion of the helices into lipid bilayer. This membrane binding leads to a steadystate conformational change in saposin C as assessed by CD spectroscopy. Clearly, the middle region of saposin C (S37W) is exposed to the aqueous phase as it is accessible to aqueous quenchers. Compared to saposin C, the middle region of saposin A is within the membrane environment and at a similar depth as helices 1 and 5. A model is proposed to account for the differences in saposin A and C activities vis-a-vis acid β-glucosidase ( Fig. 7A and B). These models were derived from the non-glycosylated and recombinant saposins, and oligosaccharides attachment may alter some of the proposed interactions between protein and lipids. Since the region between helices 1 and 2 of saposin C is solvent exposed and contains the glycosylation site, the proposed model likely would not be greatly affected. However, the additional oligosaccharide moiety on saposin A is located in a region that might be expected to have unfavorable effects on its incorporation into the membrane to the depth proposed in the model by guest on October 30, 2017 http://www.jbc.org/ Downloaded from (Fig. 7B). This could make saposin A more poorly lipid-associated as observed by Vaccarro and colleagues using natural saposin A and phospholipid liposomal membranes (38).
Of the prosaposin derived saposins, saposin A shows the highest amino acid identity/similarity to saposin C. In general, the saposin fold has a structure with five amphipathic α-helices folded into a single globular domain (17,18). The presentation of the fold is along a centrally located helix at amino-terminal (helix 1), against which helices 2 and 3 are packed from one side and helices 4 and 5 from the other side (Fig. 1A). This fold has been suggested to provide the primary interface for membrane interaction. CD and FTIR analyses indicate that the lipid-bound NK-lysin retains the saposin fold and that the amino-and carboxyl-termini of NK-lysin are in proximity to the middle region (22). This region localizes to the lower half of the globular saposin fold, which may be embedded into the lipid bilayer. The other half of NK-lysin has the negatively charged residues that are exposed to solvent. Our results with saposin A/BPS complexes indicate a similar structural orientation (Fig. 7A). In comparison, quenching of saposin C (S37W) by acrylamide, supports orientation of the middle region toward the solvent environment. With saposin C, the CD spectral changes upon binding show a reorganization and reorientation of the helices in saposin fold upon membrane interaction.
Electrostatic interaction between saposins and membrane is proposed to be the first step of impact on the saposin/membrane interactions. In our assay systems, this possible impact should be same for saposin A and C, and, thus would not affect our conclusions.
Saposins A and C have highly identical/similar sequences and conserved α-helical domains.
However, saposin A is not a significant physiological activator of acid β-glucosidase. Previously, our in vitro data showed activation of acid β-glucosidase by saposin A at µM levels whereas saposin C's effects were at nM levels (15). The present studies show a direct correlation of activation functions and orientation of the saposins A and C in the phospholipid membranes.
Fluorescence and CD spectral analyses also show different BPS induced membrane conformations in these mutant saposin Cs compared to the wild-type. To evaluate the relationship of these changes to activity, mutant saposin As were generated that preserved the helical domains for activation, but the non-helical sequence, KGEMSRP, between helices 4 and 5 was mutaganized selectively to be more "saposin C-like" LEEVSP. At nmolar concentrations, saposin As (G64E) and (K63L/G64E/M66V) activated acid β-glucosidase to nearly the same levels as wild-type saposin C. The apparent K d (activation) for these mutants were decreased over 20-fold compared to wild-type saposin A, but was about twice that of wild-type saposin C.
The significant red shift in 8 EM max of these mutant saposin As indicates a dramatically increased polarity of the solvent surrounding the tryptophan residue (W37) in saposin A upon binding to lipid bilayer. These results indicate that helices 2 and 3 in the mutated saposin As are more solvent exposed, and more closely resemble those in saposin C. Thus, the solvent exposure of the region between helices 2 and 3 in saposin C is directly related to its activation potential and lack of such exposure explains the poor activity of saposin A in this assay. The results showed that the membrane-association modes of saposins A and C are the determinants of their specific biological functions. An extension of this observation implies that the diverse functions of saposins or saposin-like proteins result from differential orientations of their helices with respect to lipid membranes.       The middle region of saposin C is exposed to aqueous phase.