The Binding of Factor H to a Complex of Physiological Polyanions and C3b on Cells is Impaired in Atypical Hemolytic Uremic Syndrome¹

Red blood cells and buffers

Blood from sheep and from two healthy human adults was collected by venipuncture and the red cells were frozen at −80°C (42). The University of Texas Health Science Center institutional review board approved protocols, and written informed consent was obtained from all human donors. The buffers used were: PBS (10 mM sodium phosphate, 140 mM NaCl, 0.02% NaN₃, pH 7.4); veronal buffered saline (VBS), 5 mM veronal, 145 mM NaCl, 0.02% NaN₃, pH 7.3; GVB, VBS containing 0.1% gelatin; GVBE, GVB containing 10 mM ethylenediaminetetraacetic acid (EDTA); DGVB, half-physiological ionic-strength buffer prepared by diluting GVB two-fold with 5% dextrose in water; MgEGTA, 0.1 M MgCl₂, 0.1 M EGTA (ethyleneglycoltetraacetic acid), pH 7.3.

Purified proteins

Complement proteins fH (43), C3 (44, 45), factor B (46), and factor D (47) were purified from normal human plasma as described. A fragment consisting of C-terminal residues 1107-1231 of fH (domains 19-20, i.e. rH19-20) was cloned, expressed and purified as described (41). Site directed mutagenesis was used to generate sixteen mutant versions of rH19-20, using the Quickchange^R mutagenesis system (Stratagene, Cedar Creek, TX). Ten of these are associated with aHUS (D1119G, R1182S, W1183R, T1184R, L1189F, L1189R, S1191L, R1210C, R1215G, and double mutant S1191L;V1197A) (48); the remaining six mutants (E1172R, K1188Q, R1203S, R1203Y, R1210S, and K1230A) have not so far been observed in aHUS patients, but were generated to probe C3b- and polyanion-binding sites. The residues chosen for the design of these mutants lie in putative C3b- and polyanion- binding sites (40, 41) and cover a wider surface region within domain 20, which is where most aHUS-associated mutations in fH lie. The proteins were cloned and expressed using the Pichia pastoris strain KM71H as previously described (41). Mass spectra of recombinant proteins (purified by cation-exchange chromatography and stored at −80°C in PBS as described previously (41)) were consistent with predicted molecular weights. A ¹⁵N-labelled sample of each mutant was prepared by using (¹⁵NH₄)₂SO₄ (Isotec) as the sole source of nitrogen during cell growth. Protein concentrations were determined (280 nm) using theoretical E^1%_{1 cm} values (49). Factor H and an aliquot of wildtype rH19-20 (100 μg) were radiolabeled with ¹²⁵I to specific activities of 4 to 6 μCi/μg as described (35).

Nuclear magnetic resonance (NMR) methods

For NMR, samples of ¹⁵N labeled proteins were 100 μM in 20 mM deuterated sodium acetate, 10% D₂O, pH 4.5. In each case a [¹⁵N,¹H] heteronuclear single quantum coherence (HSQC) spectrum was collected at 37°C on a Bruker AVANCE 800 MHz spectrometer fitted with a 5 mm CPDCI cryoprobe with Z gradients. Based on existing spectral assignments for wild type rH19-20 (41), analyses were conducted to reveal any changes in the ¹⁵N and ¹H chemical shifts of each backbone amide consequent upon mutagenesis.

Heparin affinity chromatography

Protein samples (3 μM, 1 ml) in 20 mM potassium phosphate buffer, pH 7.4, at 20°C, were loaded individually onto a HiTrap heparin-affinity chromatography column (7 mm × 25 mm, GE Healthcare, Piscataway, NJ), equilibrated with the same buffer, and subsequently eluted with a 20-column volume gradient from 0–1 M NaCl in 20 mM potassium phosphate buffer, pH 7.4. Protein elution was monitored using absorbance at 280 nm. Similar experiments were carried out for the aHUS-associated mutants (0.6 μM, 5 ml) in 20 mM HEPES, pH 7.0, at 4°C, and eluted with a 20 column volume gradient from 0-1 M NaCl in 20 mM HEPES, pH 7.0.

Gel-mobility shift assays (GMSA)

Oligosaccharides were prepared for GMSA from low-molecular weight heparin by partial digestion with heparinase I followed by size-fractionation on a BioGel P10 gel-filtration column (BioRad, Hercules, CA) (50). In the case of the tetrasaccharide, dp4, strong anion-exchange chromatography (AS-17 Ionpac, Dionex, Sunnyvale, CA) was subsequently performed to isolate only fully sulfated product (ΔUA2S-GlcNS6S-IdoA2S-GlcNS6S). Due to the heterogeneity associated with the length of the hexasaccharide (dp6) this purification step was omitted in dp6 preparations resulting in variable sulfation patterns. Fluorophore-labeled species were produced by attachment of 2-aminoacridone to the oligosaccharide reducing end (50), and GMSAs were performed, as described previously (51).

Analysis of C3b binding by surface plasmon resonance (SPR)

A BIAcore 3000 instrument was used to perform SPR. Human C3b (800-2000 response units (RU)) was amine coupled to a CM5 sensor chip (GE Healthcare). In addition, a streptavidin (SA) sensor chip (GE Healthcare) was coated with biotinylated C3b, generated as recently described (52). For each type of sensor chip, three flow cells (Fc2, Fc3 and Fc4) were independently coupled with C3b. Then 20 μl injections of the rH19-20 mutants and wildtype rH19-20 were performed at 0-12 μM over the flow cells at 5 μl/min, at 22°C. The running and sample buffer consisted of PBS with 0.005% Tween. The maximum equilibrium binding of the proteins to C3b was measured by subtracting RUs obtained for the reference cell (Fc1) from the C3b-coupled flow cell RUs. The equilibrium binding values for each concentration were plotted using the Grafit 5 program (Erithacus), which determined the apparent K_d by fitting the data to a single-site saturation curve. Since the wildtype rH19-20 interacts with C3b with a 1:1 stoichiometry, the K_d values of the mutant proteins were calculated by setting the capacity (as a constant) to that of the average (n=12) capacity of wildtype rH19-20, for each C3b-coupled sensor chip.

Binding assays with C3b-coated cells and zymosan

Sheep erythrocytes (E_SC3b) bearing a total of 1 μg C3b were incubated with ~20 ng of radioiodinated human fH and 0 to 8 μM non-labeled rH19-20 (wildtype or mutants) in 100 μl half-ionic strength buffer (DGVB). Alternatively, the cells were incubated with ~20 ng ¹²⁵I-rH19-20 and the non-labeled proteins. After 20 min at 22°C, the bound and free radiolabeled proteins were separated as described (34). The competition experiments with C3b-coated zymosan (ZymC3b) were carried out as described above, except only the IC₅₀ dose of the unlabeled wildtype rH19-20 (1 μM) was compared to a 1 μM dose of various rH19-20 mutants.

Hemolysis assays

Lysis of human erythrocytes (35) (E_H) was measured by pre-incubating 5 × 10⁶ E_H with 7 μg/ml anti-CD59 monoclonal antibody (clone MEM43; Abcam, Cambridge, MA) for 20 min at 4°C, and then mixing, on ice, with GVB, normal human serum (NHS; 40% final), and 0.1 M MgEGTA (5 mM final concentration) in the presence of 1 μM wildtype or mutant rH19-20. This concentration of wildtype rH19-20 is enough to achieve ~50% hemolysis (IC₅₀) of anti-CD59 treated cells, due to inhibition of fH-mediated protection. The mix was immediately transferred to a 37°C water bath and incubated for 20 min. To determine the extent of hemolysis induced by the mutants versus the wildtype rH19-20, 200 μl cold GVBE was added, the samples were centrifuged, and the optical density of the supernatant was determined at 414 nm. The percent lysis was determined by subtracting the A₄₁₄ of the background lysis control, and dividing by the maximum lysis in water.

Statistical analysis

Conventional Student's t tests (2-tailed) were used to assess the statistical significance of the difference between wildtype rH19-20 and the mutants in their ability to bind to C3b (BIAcore), to C3b on zymosan, and to lyse E_H in the presence of NHS and p values <0.05 were considered statistically significant.

A panel of mutant recombinant forms of fH domains 19-20 (rH19-20) was expressed in P. pastoris

A total of sixteen mutations of rH19-20 were expressed in P. pastoris, as summarized in Figure 1A. Ten of these correspond to aHUS-linked sequence variants (38). As previously noted (41) these, to a large extent, congregate on one face of module 20 (Fig. 1A). A further six mutants, that have not been identified in aHUS patients, but that cover a wider surface region of the disease-linked domain 20, were designed to help delineate binding sites for C3b and heparin. Based on previous studies (40, 41), the residues chosen for these “designer” mutants are hypothesized to lie within C3b- or polyanion-binding sites. All the mutants were homogeneous as judged by SDS-PAGE under both oxidizing and reducing conditions (not shown). The expected molecular weight of all the mutants was confirmed by electrospray ionization mass spectrometry (not shown).

The aHUS-linked mutations do not perturb structure

Samples of wildtype-sequence rH19-20, and of the mutant forms, were isotopically enriched with ¹⁵N. Each sample was used to collect a 2D ¹⁵N,¹H-HSQC spectrum. Nearly every amino acid residue within wildtype rH19-20 yielded a distinct peak in such a spectrum, as reported previously (41) and as expected for a properly folded polypeptide. For fifteen out of sixteen mutants, the ¹⁵N,¹H-HSQC spectra (not shown) were also consistent with folded proteins; the exception was the designed mutant R1203Y, which was judged to be unfolded by ¹H NMR. Furthermore, a comparison of HSQC spectra showed good conservation of chemical shifts with variations confined to the vicinity of the mutated residues. Moreover, lineshapes and spectral quality were consistent with monomeric protein for all proteins tested. These observations suggest that none of the aHUS-linked mutations resulted in major structural changes.

The aHUS-linked mutants bind with varying affinity to heparin

Previous work suggests that host GAGs form at least part of the target recognized by the C-terminal anchoring region of fH (25, 28, 31, 35, 37, 41). Thus it was of interest to assess the extent to which our mutations affect GAG binding. Heparin is commonly used as an experimental model compound for GAGs that are found on cell surfaces such as heparan sulfate and dermatan sulfate. The wildtype and mutant forms of rH19-20 were loaded onto a heparin-affinity column and eluted using a salt gradient (Fig. 2A-B). Some aHUS-linked mutants bound to the resin more tightly, others bound more weakly and the remainder eluted similarly to wildtype rH19-20. For example, the strongest interaction was found with W1183R (eluting at 520 mM NaCl), while the weakest was R1182S (eluting at 325 mM NaCl). The wildtype rH19-20 eluted at 370 mM NaCl. In general, mutations that resulted in gain of positive charge increased affinity (aHUS mutants W1183R, T1184R, L1189R; designed mutant E1172R), while mutations causing a loss of positive charge decreased affinity (aHUS mutants R1182S, R1210C, R1215G; designed mutants K1188Q, R1203A, R1210S, K1230A). To investigate further, GMSAs were performed. In these assays, the migration of a fluorescently tagged defined-length GAG fragment through an agarose gel was monitored in the presence of wildtype rH19-20 or the mutants. The wildtype protein binds and hence retards the mobility of both a chemically pure fully sulfated heparin-derived tetrasaccharide (dp4, data not shown), as well as heparin-derived hexasaccharides (dp6) that contain heterogeneous sulfation patterns (Fig. 2C). The experimental conditions were optimized so that the wild type showed some free dp6 migrating towards the anode. Smaller amounts of observed free dp6 for some mutants indicated stronger binding, while larger amounts corresponded to weaker binders. The ranking of affinities of rH19-20 mutants (aHUS-associated and designed mutants) for GAGs, as inferred from the GMSA (Fig. 2C), was very similar to the ranking derived from heparin-affinity chromatography (Fig. 2A-B). In summary, there is no uniform link between the affinity for heparin of the aHUS-associated mutants and aHUS pathology. Thus, some of the aHUS-associated mutants (W1183R, T1184R, L1189R, L1189F, S1191L, and S1191L;V1197A) were either unaffected or had increased affinity for heparin.

The aHUS-linked mutants bind with varying affinity to C3b

We measured the affinity of each mutant protein for C3b using SPR. C3b was immobilized on a CM5 sensor chip by amine coupling. Figure 3A summarizes the value of K_d obtained for each mutant. Four of the aHUS mutants (D1119G, R1182S, R1210C, and R1215G) bound with lower affinity to immobilized C3b, compared to the wildtype. The other aHUS mutants were either unaffected in their ability to bind (V1197A in the context of the S1191L;V1197A double mutant) or had a slightly (L1189F) or significantly (W1183R, T1184R, L1189R, S1191L) increased affinity for C3b. It is noteworthy that most of the aHUS mutants that clearly bind more strongly to C3b (i.e. W1183R, T1184R, L1189R) correspond to those with a higher affinity for heparin, while most of those with low affinity for C3b (i.e. R1182S, R1210C, R1215G) correspond to those exhibiting weaker heparin binding, as summarized in Figure 4. Interestingly, the aHUS mutant S1191L, which bound with greater affinity to C3b (Fig. 3-4), maintained a normal ability to bind heparin (Figs. 2 and 4).

A second type of sensor chip, pre-immobilized with streptavidin (SA chip), was used to immobilize C3b molecules that had been biotinylated at the cysteine-SH within the thioester site. This mode of immobilization emulates more closely the natural orientation of C3b on surfaces (52). Figure 3C shows representative equilibrium binding single-site saturation curves for rH19-20 wildtype as well as examples of a high-affinity (L1189R) and a low-affinity (R1215G) mutant binding to C3b. The K_d values obtained with the SA-C3b sensor chip (Fig. 3B) were very similar to those obtained on the CM5 chip (Fig. 3A). These findings indicate that the mode of C3b coupling did not bias the measured K_d values.

In order to further investigate the unexpected increase in affinity for C3b of some mutants as observed by SPR, C3b-coated zymosan particles (ZymC3b; which are devoid of polyanions) were incubated with radiolabeled wildtype rH19-20 in combination with unlabeled wildtype or the mutants with the highest and lowest affinity for C3b (by SPR). Figure 3D shows that aHUS mutants W1183R, T1184R, and L1189R as well as designed mutant E1172R, competed with radiolabeled rH19-20 more efficiently for binding to surface-bound native C3b than the unlabeled wildtype rH19-20, while aHUS mutants R1182S and D1119G were less efficient. The differences between the ability of the wildtype versus each of the mutants to bind to ZymC3b were all significant (p<0.05). These results agree with the SPR data, confirming that some disease-linked mutants do indeed bind with higher affinity to C3b. Taken together, the C3b-binding data show that there is no correlation between affinity for C3b and the association of the mutations with aHUS.

Delineating C3b and heparin binding sites

Considering our panel of sixteen rH19-20 mutations as a whole, fourteen neither introduced a cysteine (the exception being R1210C) nor resulted in structural disruption (the exception being R1203Y). Of these fourteen, nine had a positive or negative effect, compared to wildtype rH19-20, on both heparin and C3b binding, while one (V1197A, within the context of S1191L;V1197A double mutant) had no significant effect (Fig. 4). Two other mutations (R1210S and K1230A) decreased heparin binding, but did not alter C3b binding. Mutant D1119G significantly decreased C3b binding, but had no effect on heparin binding. On the other hand, S1191L increased C3b binding, while having no effect on heparin binding. Since these mutations resulted in only localized structural changes it seems valid to summarize their various functional consequences on the 3D structure of wildtype rH19-20 (Fig. 1B) and to infer that the wildtype residue is directly involved in ligand-binding if its substitution alters affinity. On this basis, Figure 1B illustrates that there is an overlapping binding patch on one face of module 20 for heparin and C3b. Two of the mutants that had minor or negligible effect on heparin binding (L1189F and V1197A, respectively) both involve conservative substitutions of buried side chains.

The rH19-20 aHUS-associated mutants are impaired in their ability to bind to C3b-coated host-like erythrocyte surfaces

A clear message of the results described above is that while ten mutants included in this study have been linked with occurrence of aHUS (48), not all of them are negatively affected in their ability to bind to C3b and polyanion ligands (summarized in Fig. 4). We therefore tested whether these mutations interfere with the ability of fH to bind, simultaneously, to C3b and host-like polyanions in a more physiological setting using a competitive binding assay, as previously described (35). In this assay, rH19-20 competes with radiolabeled full-length fH, inhibiting its binding to C3b and host-like polyanions on sheep erythrocytes (E_S), which carry high levels of sialic acid-containing polysaccharides (33, 34) (as do most human cells and tissues). We showed previously that such competition impairs fH complement-regulatory functions and increases complement activation on host surfaces (35). Thus, the C-terminal module-pair is important for anchoring fH to self-surfaces following C3b deposition. Data in Figure 5A confirms that wildtype-sequence rH19-20 inhibits, in a dose dependent manner, the binding of full-length fH to C3b-coated host-like cells, with an IC₅₀ of 0.2 μM. The aHUS-linked mutant forms of rH19-20, on the other hand, were significantly impaired in their ability to compete, displaying from < 2.5% (R1215G, D1119G and R1182S) to 40-50% (T1184R and R1210C) of the inhibitory activity of the wildtype rH19-20 (Fig. 5A). Only the designed mutant (E1172R) retained more than 70% activity in this assay. The presence of the V1197A mutation within the double mutant S1191L;V1197A did not further impair its ability to compete versus S1191L alone.

We also determined the effect of the mutations on the ability of rH19-20 to compete with radiolabeled wildtype rH19-20, i.e. without the potential interference of the other C3b/polyanion-binding sites or higher-order architecture of full-length fH, and very similar results were obtained (Fig. 5B). Thus, the four aHUS mutants that were most affected (D1119G, R1182S, W1183R, and R1215G) and the two least affected mutants (T1184R and R1210C) in the competition assay with full-length ¹²⁵I-fH, ranked comparably in the competition assay with ¹²⁵I-rH19-20, and can be compared in summary Figure 4.

Ability of rH19-20 aHUS-associated mutants to inhibit fH-mediated protection of human erythrocytes from complement-mediated lysis

Experiments were performed in which complement-mediated lysis of human erythrocytes was measured in the presence of normal human serum and rH19-20 mutants (Fig. 6). Wildtype rH19-20 competes with fH, inhibiting fH-mediated protective action more effectively than any of the aHUS-linked mutants, apart from R1210C and T1184R, which are comparably effective. This is largely consistent with the cell-surface binding results (Fig. 5) and provides further evidence that anchoring of fH to self-surfaces via its C-terminus is essential for its protective function.

Thus, the competition data strongly indicate that a common feature of aHUS-linked fH mutants is the reduced ability to engage, in a functionally effective way, with the fH target (C3b/C3d), when in the context of polyanion-rich self-surfaces. Intriguingly, this behavior on physiological cell surfaces does not correlate with the increased affinity that some of the aHUS mutants displayed for C3b (Fig. 3) and/or for the GAG-model compound heparin (Fig. 2), when these interactions were analyzed individually.