| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
*
Department of Biochemistry, University of Texas Health Science Center, Tyler, TX 75708; and
Complement Research Unit, Haartman Institute, University of Helsinki, Helsinki, Finland
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The alternative pathway of complement appears to activate spontaneously. It continuously deposits C3b on all particles in contact with blood (reviewed in Ref. 31). Following covalent attachment of C3b to a potential target a control system either inactivates the C3b or allows it to initiate an amplification system that is capable of covalently attaching millions of C3b molecules to the target surface in <1 h (32, 33). Three mechanisms of regulation of this inactivation/amplification process have now been identified. Two of these operate in a manner different from most innate immune systems; they do not directly recognize the pathogen/Ag, but instead prevent attack on the host and allow attack on everything else. This design is sophisticated, in that it can adapt to new organisms as they arise unless they become host-like (reviewed in Refs. 22 , 27 , and 30). First, membrane-bound regulators protect host cells from attack by the APC by inactivating complement enzymes and proteins on host cell surfaces (reviewed in Refs. 5 , 34 , and 35). Second, and the subject of this paper, preventing amplification of the initial C3b also depends on the affinity between the plasma complement control protein factor H and the bound C3b. The affinity between C3b and factor H is regulated by interactions among functional sites on factor H, sites on C3b, and structures on the surface to which C3b is bound. The present work demonstrates that each of the numerous sites on factor H plays a distinct role, and their involvement is different on different biological particles. The third mechanism for regulating activation of the alternative pathway is carbohydrate specific. Once C3b amplification is allowed to begin due to the lack of membrane regulators or weak factor H binding, the efficiency of C3b attachment to the surface becomes important. We have shown that amplification is dependent on the specificity of nascent C3b for different carbohydrates (6). C3b attachment to polysaccharides exhibiting low reactivity with the thioester site is inefficient, and this inefficiency affects each round of amplification. The end result is that activation of the APC (i.e., amplification) can be weak or strong depending on the specificity of metastable C3b for the particular carbohydrate structures on that particle (6, 12, 19, 26, 36).
Factor H is composed of 20 homologous domains, each containing 
60 aa
(37). Numerous functional sites have been identified along
this structure (38, 39, 40, 41, 42, 43, 44, 45). Three of these sites have been
shown to interact with unique sites on the C3b molecule (39, 44, 46, 47). The C3b binding site at the N-terminus possesses
decay-accelerating activity for the alternative pathway C3/C5
convertase and also serves as a cofactor site for factor I, a serine
protease that inactivates C3b. At least three polyanion binding sites
are located along the factor H molecule. Interactions between factor H
and surface polyanions have been shown to be important to
discrimination between activators and nonactivators of the alternative
pathway of complement (7, 8, 11, 12, 14, 15, 17, 18, 23, 24, 27, 29, 43, 45, 48, 49, 50). This paper presents an analysis of the
influence that different regions of factor H have over
decay-accelerating activity and affinity for C3b. We have examined
these interactions with C3b bound to four different cell types (two
activators and two nonactivators). The results indicate that different
functions of factor H rely to varying degrees on multiple sites along
the factor H molecule and that the reliance on individual sites varies
depending on features on the target surface recognized by this
protein.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
C3 (51, 52), factor B (53), factor D (54), and factor H (55) were purified from normal human plasma as described previously. Complement component C3 was repurified on a Mono S column (Pharmacia, Piscataway, NJ) (52). Cobra venom factor and cobra venom factor,Bb were prepared as previously described (56). All proteins were stored at -75°C in VBS. The concentrations of factor H and all rH proteins were determined at 280 nm using E1%1 cm of 12.4. Zymosan, Trizma base, glycine, guanidine hydrochloride, and polyethylene glycol were purchased from Sigma (St. Louis, MO). Linearized Baculogold baculovirus DNA was purchased from PharMingen (San Diego, CA). The Bac-To-Bac baculovirus expression system, Escherichia coli strain DH10Bac, Grace’s insect medium, yeastolate extract, lactalbumin hydrolysate, sodium bicarbonate, and gentamicin were purchased from Life Technologies (Gaithersburg, MD). E. coli strains NM 522 and Topp 10' as well as Spodoptera frugiperda (Sf9) and High Five cells were purchased from Invitrogen (San Diego, CA). FCS was purchased from Atlanta Biologicals (Norcross, GA). Restriction enzymes, T4 DNA ligase, and calf intestinal alkaline phosphatase were purchased from New England Biolabs (Beverly, MA). GeneAmp PCR reagents were purchased from Perkin-Elmer (Norwalk, CT). Buffers used were veronal-buffered saline (VBS), 5 mM veronal, 145 mM NaCl, and 0.02% NaN3, pH 7.3; VBE, VBS containing 10 mM EDTA; GVB, VBS containing 0.1% gelatin; GVBE, GVB containing 10 mM EDTA; DGVB, half-physiological ionic strength buffer prepared by diluting GVB 2-fold with 5% dextrose in water; and PBS, 10 mM sodium phosphate and 140 mM NaCl, pH 7.4.
Preparation and purification of recombinant proteins
Site-specific deletions in human factor H cDNA (57) were created by overlap extension PCR (58) as previously described (44). The constructs were restriction digested with XbaI and ligated with XbaI-treated, dephosphorylated pBacPAK 8/9 (Clontech, Palo Alto, CA) or pFastBac1 (Life Technologies). The recombinant plasmid was transformed into competent Topp 10' or DH10Bac cells and checked for correct orientation by restriction mapping. All PCR-derived regions of DNA were sequenced entirely to verify that the expected mutation had been made and to determine that no unwanted alterations in amino acid sequence had been introduced. Sf9 cells were either transfected with different constructs in pBacPAK 8/9 plus Baculogold baculovirus DNA or with constructs in pFastBac1 using the Bac-To-Bac baculovirus expression system (Life Technologies) according to the manufacturers’ procedures. Recombinant viruses developed using Baculogold DNA were further cloned by end-point dilution. The Sf9 cells were maintained in complete insect medium containing 10% FCS and 50 µg/ml gentamicin at 27°C. Cells in suspension culture were maintained in Spinner flasks at 80 rpm on a magnetic stirrer. Media were collected after 4 days and analyzed by Western blotting for synthesis of recombinant proteins. Polyethylene glycol was added to media to a final concentration of 15%, and the mixture was stirred at 4°C for 30 min. The precipitated proteins were collected by centrifugation, redissolved in VBE, and applied to an anti-factor H-Sepharose immunoadsorbant column. Recombinant factor H was eluted with 6 M guanidine, dialyzed against VBS, and concentrated by ultrafiltration.
Radiolabeling
Human factor H and recombinant H proteins (20 µg in 20–140 µl) were labeled with 50 µCi of 125I for 30 min at 0°C in a glass tube coated with Iodogen (Pierce, Rockford, IL). After the incubation 250 µg of potassium iodide and 500 µg of OVA in 50 µl of buffer were added. The free 125I was removed by centrifugal desalting through G25 pre-equilibrated with GVB (59). Specific activities of the labeled proteins ranged from 0.2–2.0 µCi/µg. Specific activities for 125I-labeled factor B ranged from 0.6–1.4 µCi/µg.
Preparation of C3b-coated cells
Deposition of C3b on zymosan (Zym) and erythrocytes (sheep, ES; rabbit, ER; and human, EH) was accomplished using purified C3 and factors B and D as previously described (11, 60, 61), with the substitution of nickel for magnesium and nephritic factor (62) to stabilize the C3 convertases on the surface of the cells. The number of C3b molecules bound was determined to be between 41,000 and 170,000/particle by radiolabeled Bb binding (61).
C3/C5 convertase decay acceleration assays
Decay-accelerating activity expressed by factor H and the rH
proteins was measured by determining their ability to accelerate the
natural release of 125I-labeled Bb from
cell-bound C3b,Bb. The C3b,Bb complexes were formed by incubating
2.3–4.6 x 107 ZymC3b,
ESC3b, ERC3b, or
EHC3b with 0.9 µg (0.5 µCi) of
[125I]factor B and 0.6 µg of factor D in 75
µl of GVB containing 1 mM NiCl2 at 22°C for 3
min. EHC3b were preincubated with sufficient
rabbit anti-DAF purified IgG (47 µg) to inactivate all the
intrinsic DAF activity of the cells (63). Anti-DAF-treated
EHC3b exhibited the same Bb decay rate as cells
lacking DAF. Formation of the C3 convertase was stopped by the addition
of 145 µl of GVBE. The cells (20 µl) were added immediately to
reaction mixtures containing varying amounts of human H, rH, or a rH
mutant (0–10 µg) in 80 µl of GVB. After 10 min at 22°C the cells
were sedimented rapidly (1 min, 10,000 x g) through
250 µl of 20% sucrose in GVB in a microfuge tube. The bottoms of the
tubes were cut off, and the radioactivity in the cell pellet and the
supernatant was measured to determine the percentage of Bb remaining
bound. The specific activity of factor H-derived proteins in this assay
was determined from the concentration required to release 50% of the
[125I]Bb compared with that remaining bound in
the absence of factor H.
Binding assays
C3b-bearing cells (ZymC3b, ESC3b, ERC3b, and EHC3b) had between 41,000 and 170,000 C3b/cell. Varying amounts of cells were incubated for 15 min at 22°C with 10–20 ng of radiolabeled humH, rH, or mutant rH proteins in half-ionic strength buffer (DGVB) in a total volume of 100 µl. Bound and free radiolabeled proteins were separated by layering 80 µl of the mixture on top of 20% sucrose in DGVB, centrifuging for 2 min at 10,000 x g, and cutting the tube to separate the pellet and supernatant (11). Maximum functional radiolabeled protein was determined for each preparation using saturating amounts of C3b (up to 300 x 1011 C3b/assay). The percentage bound is reported minus nonspecific background binding to the appropriate cell lacking C3b. This was, in all cases, <2% of the total counts per minute.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Studies of the functional roles of different sites on factor H
employed proteins containing five domain and 10 domain deletions (Fig. 1
). The strategy used (44)
resulted in exact deletion of whole domains starting from the first Cys
of each domain and ending with the residue before the first Cys of the
next expressed domain. This strategy removes the domains as well as the
entire interdomain linker on the C-terminal side of the deleted
domains. The proteins were produced in a baculovirus expression system,
purified by immunoaffinity chromatography (44), and stored
frozen at -75°C before analysis.
|
The cell surface to which a C3/C5 convertase is bound is known to
determine the effectiveness of factor H decay activity (11, 15, 48). As a consequence, any examination of the roles of different
sites along the molecule must include particles representative of
activators and nonactivators of the human alternative pathway of
complement. We have chosen two representatives from each group.
Nonactivators were represented by a host cell (human erythrocytes,
EH) and by sheep erythrocytes
(ES), the traditional nonactivator for complement
studies. Activators were represented by a similar cell type, an
erythrocyte, but one that is a strong activator (rabbit erythrocytes,
ER) and by a traditional complement activator of
microbial origin (yeast cell walls or Zym). C3b was deposited on the
cells using purified complement components, and the C3/C5 convertase
(C3b,Bb) was formed on the surface using factor D,
Ni+2 and radiolabeled factor B. The factor
H-mediated decay of this C3b,Bb enzyme was measured by determining the
radiolabel remaining bound after incubation for various times in the
presence and the absence of factor H, as shown in Fig. 2. The specific activities of human
factor H and the different recombinant factor H mutants were determined
by measuring the percentage of [125I]Bb
remaining bound after 10 min at 22°C compared with that remaining
bound in the absence of any factor H. Plots of the concentration
dependence of decay acceleration (Fig. 3)
were used to calculate the molar concentration needed to decay
dissociate 50% of the C3/C5 convertases during the 10-min incubation.
Comparison of the molar specific activities of eight species of factor
H on four different cell types is shown in Table I. The values were normalized to the
molar specific activity of wild-type human factor H on
ESC3b,Bb cells to aid comparison. In these assays
only 48 fmol of factor H (480 pM) was required to decay accelerate 50%
of the C3/C5 convertase on ESC3b. The full-length
rH and human factor H exhibited almost identical specific activities on
each cell type (Table I). All recombinant proteins that lacked the
first four domains of the protein (rH
1–5 and rH1–10) also
lacked decay-accelerating activity. Trace levels of factor H activity
were observed in sham purifications from fresh medium containing FCS
and account for the trace (<0.02%) background activity in Fig. 3.
Activities of the other deletion mutants were very similar on the two
nonactivators, ES and EH.
The intrinsic decay-accelerating activity present on human erythrocytes
due to DAF (CD55) was inactivated by including sufficient purified
rabbit anti-DAF to inactivate all the DAF (63). The
decay rate of the anti-DAF-treated EHC3b,Bb
was indistinguishable from that of ESC3b,Bb.
|
|
|
and Table I). Deletion of the
second C3b binding site, localized to the SCR6–10 region in
our previous study (44), showed little effect on decay
acceleration, except on ER. Removal of the
polyanion binding site located in the 11–15 domain region of factor H
(40) showed little or no effect on decay acceleration with
the two activators, as expected. On the sialic acid-bearing
ES and EH, however, there
was a 6- to 10-fold drop in specific activity with the loss of this
region. Similarly, removal of the sialic acid/polyanion binding site
and the third C3b binding site located at the C-terminal of factor H
(rH16–20) had almost no effect on the activators, while
resulting in an 11- to 16-fold drop in activity on nonactivators (Table I). Deletion of the entire C-terminal half of the molecule
(rH11–20) removed both the polyanion binding sites and
the C3b binding site and resulted in the loss of 97% of the
decay-accelerating activity on ES and
EH, indicating that the sites in this region of
the protein are important for full expression of decay-accelerating
function on host cells and other sialic acid-bearing cells. On Zym,
loss of this region had no effect. In fact, the C-terminal 15 domains
of factor H appear to play no role in decay acceleration when the C3/C5
convertase resides on zymosan. Thus, the high efficiency of convertase
decay on host cells and other nonactivators is not the result of
interactions at the decay-accelerating site itself, but is regulated by
interactions with domains of factor H that are perhaps >800 Å away
(64, 65, 66). Binding of factor H: cell surface-dependent contributions of different domains
Our previous study measured binding of rH deletion mutants to
ESC3b (44). The present study
expands the examination of binding to investigate the influence of the
cell surface to which C3b is attached. Each of the purified recombinant
proteins was radiolabeled, and binding was measured to the various
C3b-loaded cells. Because the rH proteins were of limited supply, the
C3b concentration, instead of factor H, was varied to generate binding
curves (Fig. 4). In Fig. 4, A
and B, it can be seen that a maximum of 79% of the
radiolabeled human H and 60% of the radiolabeled rH were capable of
binding to ESC3b. The maximum binding of
rH11–15 was 11%. In all other functional assays
except binding assays, which were performed after radiolabeling, the
proteins appeared to be fully active. Radiolabeled proteins had reduced
activity, and rebinding experiments on the unbound fractions indicated
that the unbound material was not capable of binding. Thus, it was
necessary to determine the amount of functional protein in each
preparation. Increasing concentrations of cell-bound C3b (as much as 10
times the highest levels shown in Fig. 4) were used to obtain maximum
binding levels for each radiolabeled protein. The maximum binding
observed with any cell type was taken as the amount of functional
protein present.
|
The binding behavior of full-length factor H was similar on the two
nonactivators and on the two activators (Fig. 4, A and
B). Like factor H, rH showed lower affinities for C3b on the
activators compared with C3b on the nonactivators. The difference is
identical with the published 10-fold affinity difference in human H on
these surfaces (12). Because rH bound to every cell type
with a 5-fold higher affinity than human H, the binding of all mutant
rH molecules will be compared with that of rH. Loss of
SCR1–5 containing the first C3b binding site (Fig. 4C) reduced the affinity for C3b on all cells, but a
significant difference was found in the affinities on the two
activators. Loss of the second C3b binding site and the first sialic
acid/heparin binding site by deletion of SCR6–10 (Fig. 4D) caused a greater reduction in affinity than loss of the
first C3b site on all four cells. However, the affinity loss on Zym was
considerably greater than that on any other cell (compare Fig. 4, C and D). This observation suggests
that the SCR6–10 region plays a greater role in the
interaction between factor H and ZymC3b than on the other cells.
Removal of the SCR11–15 region containing the second sialic
acid/heparin binding site caused a minor (2- to 3-fold) drop in the
affinity for C3b on the nonactivators ES and
EH, but no change in binding to the activators
(Fig. 4E). Finally, removal of the third C3b binding site
and the third sialic acid/heparin site (rH16–20) resulted
in the most unexpected pattern. While loss of this site dramatically
reduced the affinity for C3b on ER, it had no
effect on the other activator Zym, and while it caused a 5-fold effect
on binding with EH, loss of this region caused
the largest single-site effect on binding to C3b on
ES seen in this study. This switch in
activator/nonactivator discrimination can be clearly seen by comparing
the patterns of the two activators in Fig. 4F (filled
symbols) with patterns in the five other panels of Fig. 4.
Although the binding curves in Fig. 4 allow comparisons of relative
affinities on different types of cells, the presentation of binding
data shown in Fig. 5 makes it possible to
compare the affinities of different deletion mutants on a single type
of cell. The data have been normalized to the maximal activity of
individual radiolabeled protein preparations to correct for the
differences in functional activity of each labeled protein. In the
upper panel the binding of rH and that of
rH11–15 to ERC3b are similar (at
the 20% bound level), while the affinities of rH1–5,
rH6–10, and rH16–20 were 10-fold reduced
from that of the complete rH molecule. On ZymC3b (Fig. 5, lower
panel) the loss of the SCR6–10 region caused a 32-fold
drop in binding affinity, indicating that binding to zymosan relied
primarily on this site. The only other site involved in binding to C3b
on zymosan was the first C3b binding site in SCR1–5.
Although the effect of removing SCR1–5 was similar on the
two activators, the loss of SCR16–20 was very different. On
ERC3b the mutant rH16–20 bound
9-fold more poorly than rH, while on ZymC3b the loss of
SCR16–20 had no effect.
|
the relative differences in
apparent affinity are presented. All the values for binding affinity
are reported relative to the concentration of C3b on
ESC3b needed to bind 20% of the full-length rH
protein. In these experiments this required sufficient
ESC3b cells to provide 220 pM C3b. For
ERC3b a 10-fold higher concentration of C3b was
required to bind 20% of the protein, indicating the 10-fold reduction
in affinity for C3b bound to this cell. Table II includes data from 28
binding curves, including those for the half-molecules
rH1–10 and rH11–20, which are not shown in
Fig. 4 or 5 (data in Table II are summarized from two or more
independently measured binding curves for each cell/rH combination).
Not shown in Table II are the comparable values for normal human factor
H, which are 5, 4, 57, and 55 for ES,
EH, ER, and Zym,
respectively, relative to 1 for the rH and ESC3b
pair. The origin of this 5-fold difference has been attributed to
charge differences due to the lack of sialic acids on the insect
cell-derived rH and is illustrated by comparing Fig. 4, A
and B.
|
allow a quantitative comparison of the relative
affinities of the different deletion mutants for clusters of C3b on
four different cells. Two patterns are apparent from this table. First,
major differences in the roles of various regions of factor H were
found between activators and nonactivators of the APC. Second, there
were also significant differences between the members of these groups.
The interaction between factor H and C3b on ES
and EH appears to use the SCR1–5
region equally and differ only slightly, if at all, in their reliance
on SCR6–10. However, binding to C3b on the two nonactivators
differs considerably in the use of the C-terminal domains 16–20, where
the third C3b site and a sialic acid/polyanion binding sites are
located. In fact, on ES a 50-fold reduction in
affinity occurred with the loss of SCR domains 16–20. In contrast, on
EH the loss of this region had only a 5-fold
effect, while loss of SCR domains 6–10, containing the second C3b site
and the first sialic acid/polyanion binding site, had the greatest
effect on binding.
Comparison of the two activators demonstrates that a distinct pattern
of functional site usage governs the interactions between factor H and
C3b on these surfaces compared with the nonactivators (Table II). The
data show that activators relied more on the C3b binding site in the
SCR6–10 region than on any other site. This was most
significant with ZymC3b where there was an almost 10-fold greater
reliance on the second C3b binding site in the SCR6–10
region than on the first C3b binding site in SCR1–5 where
both decay-accelerating and factor I cofactor activity reside. Reliance
on these two sites (SCR1–5 and SCR6–10) differed
by <2-fold on ER and on the two nonactivators.
The biggest difference in site usage between the two activators was
found in the importance of the C-terminal regions. Deletion of any or
all of the last 10 domains had no effect on rH binding to C3b on Zym
(compare the 11-fold reduction in affinity found for rH (Table II) with
the 10-, 12-, and 12-fold reductions found for rH11–15,
rH16–20, and rH11–20, respectively). In
contrast, ERC3b exhibited a decrease from 10- to
90-fold in affinity for rH lacking the C-terminal SCR16–20
region. Deletion of the entire last 10 domains showed less of an effect
on ER, but data in Table II suggest that deletion
of two five-domain regions did not always have as great an effect as
would be expected from the loss of multiple binding sites.
Direct interactions between factor H and cell surface polyanions were
not detected in the present study. Although binding assays were
performed at half the normal ionic strength, binding to particles
lacking C3b was <2% for all cells. This background binding has been
subtracted from the data reported here. An additional concern regarding
ionic interactions was that the lack of sialic acids on rH, made in
insect cells, might alter the binding to cells bearing high surface
polyanion concentrations (ES and
EH) compared with cells with little surface
charge (Zym). A comparison of Fig. 4, A and B,
reveals that this is apparently not the case. Although humH bound
5-fold more poorly to ESC3b than rH, this 5-fold
difference was observed with every cell type from the highly sialic
acid-loaded EH to the nearly neutral Zym, which
lacks sialic acid. These results suggest that the primary interaction
responsible for the difference in affinity of humH and rH is due to
charge interactions between C3b and factor H.
Cross-comparison of the roles of different sites for different activities on different cells
Because the binding of full-length rH to each cell type showed
large and variable differences, it is difficult to compare the roles
that each region of factor H plays in decay acceleration and binding.
Table III allows such a comparison. Table III shows the data from both portions of this study normalized to the
activity of full-length rH on each particular cell type. This analysis
permits, for example, one to quickly see that decay acceleration by rH
on Zym was unchanged by deletion of any region of the molecule except
the decay-accelerating site in SCR1–5 (i.e., the relative
specific activity remained 1 for all deletions, except where activity
was lost due to removal of domains 1–5). This presentation of the data
also allows easy comparison of the relative importance of a region to
either activity on all four cell types. For example,
rH6–10 can be seen to be most important to decay of C3/C5
convertase activity on ER (4-fold reduction in
activity with loss of this region), but for binding this region was far
more important. The greatest effect was observed on
EH and Zym, where 19- and 32-fold reductions in
binding affinity were found. This apparent contradiction between
binding and decay-accelerating activity is also made clear by Table III
and will be discussed below. Finally, a careful study of Table III
allows a comparison of the roles of all the regions of factor H in each
activity on all cell types. What is apparent is that the functions of
factor H on every surface were dependent upon a unique set of
interactions with the different domains of factor H.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The unique structure of factor H allows it to simultaneously use
multiple recognition sites (Fig. 1). These domains appear to be
arranged as a flexible chain of beads on a string (64, 65, 66, 74), and such domains can interact with multiple ligands that
are not contiguous on target surfaces. The discriminatory ability of
such an arrangement is illustrated in the data presented in this paper.
Tables I, II, and III demonstrate that factor H interacts uniquely with
each of these four potential biological targets by using a unique set
of sites for each cell and each activity.
Expression of decay-accelerating activity for the APC C3/C5 convertase
required 65-fold higher concentrations of factor H if the enzyme was on
a Zym particle rather than on a nonactivator such as a host cell (Table I). No other sites appeared to interact with the Zym surface, because
the specific activity of rH was unaffected by removal of any region
other than the decay-accelerating site itself (Table III). Similarly,
on the second activator, ER, the C-terminal
domains of the molecule contributed only minimally to the activity of
the decay-accelerating site. On nonactivators, however, regions other
than the decay-accelerating site (SCR1–5) of factor H
enhanced activity greatly, especially the C-terminal half of
the molecule. In fact, loss of SCR domains 11–20 yields a
protein (rH11–20) that has the same specific activity as
full-length rH on Zym, but has lost all of the interactions that
enhance decay, and this mutant recognizes little difference between
activators and nonactivators (see last line of Table I). The enhanced
decay activity on nonactivators is most likely the result of
interactions with polyanions on these surfaces for the following
reasons. In terms of alternative pathway activation the major
distinguishing feature between these cells is the presence of high
densities of sialic acid on ES and
EH. The surface density of sialic acids on
ER is 10% that on EH,
and Zym has no sialic acid (12). Furthermore, we and
others have shown that removal of sialic acid from
ES and sialic acid-bearing complement-resistant
bacteria reduces the activities of factor H on these surfaces and
causes the cells to become activators of the human APC (7, 11, 12, 15, 24, 48). Interactions of multiple C3b molecules on the
surface with the C3b binding site at the C-terminal of factor H cannot
account for the observed results, because the C3b densities on the
activators and nonactivators used in this study were similar (for
example, in one set of assays the number of C3b per cell was between
76,000 and 93,000 on the four cell types and produced the same
results). Thus, the evidence suggests that the SCR11–15 and SCR16–20
regions of factor H interact with surface sialic acids to enhance
decay-accelerating activity on nonactivators of the APC.
If the involvement of individual regions of factor H in decay
acceleration is compared with the involvement of regions controlling
binding, an apparent contradiction becomes evident. For example, in
Table III the removal of domains 6–10 had no effect on decay
acceleration activity on Zym; however, this same deletion reduced
binding on Zym to 3% of that of full-length rH. Scientific principles
tell us that no biological event can occur without an interaction, thus
binding must accompany decay acceleration. If the results are correct,
then the binding occurring in these two events must be different. We
believe that the different patterns arise from the kinetic differences
between the two assays. In the assays for decay acceleration we
measured the rate of destruction of the binding site. That is, C3b,Bb
is the ligand for this binding event, and no signal occurs until it is
destroyed by interaction with factor H. Extrapolations from earlier
work (14) (M. K. Pangburn, unpublished observations)
suggest that the half-life of the C3 convertase (C3b,Bb) in plasma is
very short (in the millisecond range) due to decay by factor H. The
binding assays performed here (Fig. 4 and Tables I and III) measured a
much more stable interaction. In fact, only if the binding was stable
during the 1- to 2-min spin through 20% sucrose would a signal have
been observed in these assays. Thus, during decay acceleration the
N-terminal end of this 800-Å-long flexible protein must act like the
end of a whip that splits the C3b,Bb complex with only brief contact.
On surfaces bearing sialic acids, ionic interactions of the opposite
end of the protein with surface charges aid in making this contact.
They may also hold the protein on the surface, allowing the active end
to decay many nearby C3/C5 convertases in an 800-Å radius, thus
contributing to the enhanced effectiveness of decay on host and
host-like surfaces. Zym and most bacteria lack these sialic acid sites.
The results of binding assays suggest that a longer term interaction
also occurs that involves many other sites in factor H. In light of
these considerations, the evidence suggests that each different
function of this molecule involves a unique set of binding sites and
that their relative importance to each activity depends on the markers
on the surface to which C3b is bound.
Binding, measured here as a long-term interaction between factor H and surface-bound C3b, may be important to several activities of factor H. First, because factors B and H are competitive, with a 70-fold affinity difference in favor of factor H on ES (11), resident factor H would effectively prevent binding of factor B. With factor H bound, formation of the C3/C5 convertase (C3b,Bb) would be prevented, and APC amplification would be stopped. If factor H arrives after C3b,Bb is formed, then the enzyme would first be decay dissociated, and reformation would be prevented by the long term association of factor H with the C3b-coated surface. Second, permanent inactivation of C3b requires cleavage of C3b by the plasma serine protease factor I. Factor H is a required cofactor for factor I, and factor H, residing for extended periods bound to C3b, would provide the site necessary for factor I to function. Once cleaved, C3b no longer supports continued complement activation. Furthermore, upon inactivation to iC3b, C3b loses affinity for factor H, and the regulator would be free to move on to another C3b. Because no factor I was present during our assays, factor H remained bound.
Binding assays using the deletion mutants (Table II) indicated that the
long term interaction involves a more complex pattern of sites on
factor H than are involved in decay acceleration (Table III).
Furthermore, these multiple interactions appeared to discriminate more
strongly among all four cell types used in this study than those
regulating decay. Although decay acceleration showed little difference
between EH and ES for any
of the deletion mutants, binding showed large differences. Stable
binding to C3b on EH was more dependent on the
SCR1–10 region, while binding to ES relied less
on this region and much more on SCR16–20. In fact, a 10-fold
difference in affinity was caused by deletions at the C-terminus of the
protein on these two cells (see rH16–20, Table II). On
APC activators little difference in decay activity was observed when
SCR16–20 were removed (Table III). However, when binding was
measured, large differences were seen, with the affinity on
ER dropping 9-fold, consistent with the presence
of low amounts of sialic acids on this surface. Binding to
ES dropped 50-fold with loss of
SCR16–20, while the affinity on Zym was unaffected by this
deletion. The major site mediating binding to C3b on Zym was the second
C3b site located in SCR6–10. Clearly, this was not due to
interactions with the heparin/sialic acid binding site located in this
region (43, 44) and was most likely due to binding to the
second C3b site located in the same region. Surprisingly, binding to
C3b on human erythrocytes also depended primarily on this
SCR6–10 region and less on the N- and C-terminal five-domain
regions. Here it was less clear whether the sialic acid or the C3b
binding sites in these regions were the dominant factors. These
observations are consistent with the finding that human erythrocytes
stripped of most of their sialic acid (and all DAF and CR1 activity)
still do not activate the alternative pathway (M. K. Pangburn,
unpublished observations), whereas sheep erythrocytes stripped of most
of their sialic acid do become activators (11, 15, 48).
One caveat regarding the extension of binding data to normal human factor H is that rH exhibited 5-fold higher affinity. Even though both rH and humH had nearly identical specific activities in decay assays, and rH affinity drops 10-fold on activators in exact quantitative agreement with results for humH, some caution should be exercised as to the quantitative contributions of each site in humH. If the cause of the affinity difference between rH and humH turns out to be nonuniform across the protein, then the contribution of any affected site might be exaggerated in rH.
It is not yet clear how many functional sites factor H possesses, but its discriminatory power could be impressive. If we assume that each domain of factor H provides it with a unique ligand specificity and that these sites could work together in pairs, threes, fours, etc., then factor H would by simple combinatorial math have the ability to discriminate among >106 target surfaces. Its unique flexible structure and its length of >800 Å allow factor H the opportunity to use its 20 domains to scan a large area on a target surface for ligands. This flexibility removes the restriction that the ligands be arranged precisely on the target surface. Perhaps even more impressive, this argument assumes only two states at each site (high affinity or no affinity). If each site possesses a spectrum of affinities for different structures found on the surfaces of biological particles, then the combinatorial potential of this single molecule rapidly exceeds that of the entire Ab repertoire of the adaptive immune system. While some diversity of factor H specificities would be necessary for this innate complement system to recognize all the different cells and tissues of the host organism, there appears to be no biological need in the host for a level of diversity equal to the adaptive system to have evolved in factor H. However, micro-organisms may have adapted to the host-imposed specificities of this innate defense system and thus many appear to express varying degrees of host-like markers. Whatever drove factor H to evolve its discriminatory ability, we have begun to show how this single protein, with a defined primary structure, controls diverse target recognition in an ancient system of innate immunity. Furthermore, the molecular mechanism used by this protein to control complement activation illustrates the principles and advantages of single molecule combinatorial chemistry.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 A preliminary account of this work was presented at the 18th International Complement Workshop, Rodos, Greece, October 1998.
3 Address correspondence and reprint requests to Dr. Michael K. Pangburn, Department of Biochemistry, Biomedical Research, University of Texas Health Science Center, 11937 U.S. Highway 271, Tyler, TX 75708.
4 Current address: Department of Pathology, Uniformed Services University of Health Sciences, Bethesda, MD 20814.
5 Abbreviations used in this paper: C3b, product of proteolytic activation of complement protein C3; APC, the alternative pathway of complement; SCR, short consensus repeats that are domains 60 aa in length found in many complement control proteins; DAF, decay accelerating factor; MCP, membrane cofactor protein; humH, human factor H; rH, recombinant human factor H produced in insect cells; rH1–5, rH from which SCR domains 1–5 have been deleted; EHC3b, human erythrocytes bearing complement protein C3b on their surface; ESC3b, sheep erythrocytes with C3b; ERC3b, rabbit erythrocytes with C3b; ZymC3b, zymosan (yeast cell walls) with C3b; VBS, veronal-buffered saline; GVB, VBS containing 0.1% gelatin; GVBE, GVB containing 10 mM EDTA; DGVB, half-physiological ionic strength buffer prepared by diluting GVB 2-fold with 5% dextrose in water; Zym, zymosan.
Received for publication November 12, 1999. Accepted for publication February 22, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
) human factor H (27 µg) in a final volume of
900 µl (GVBE). The release of radioactive Bb was monitored by
removing samples at various times, layering the mixture on 20% sucrose
in GVBE, and sedimenting the cells by centrifugation at 10,000 x
g for 2 min.
, 
, 
); filled symbols, assays with ZymC3b, Bb (•, 
, 
,
). Assays shown here used the recombinant proteins rH (
