- Email: [email protected]
Isolation of a 30 kDa immunoglobulin binding protein from Pseudomonas maltophilia
Journal of Immunological Methods, 141 (1991) 187-197 © 1991 Elsevier Science Publishers B.V. 0022-1759/91/$03.50 ADONIS 002217599100244N
187
JIM06010
Isolation of a 30 kDa immunoglobulin binding protein from Pseudomonas maltophilia Sanjeev Grover, Zell A. M c G e e and William D. Odell Departments of lnternal Medicine and Physiology, University of Utah School of Medicine, Salt Lake City, UT 84132, U.S.A. (Received 12 November 1990, revised received 28 February 1991, accepted 11 April 1991)
We have demonstrated that Pseudomonas maltophifia (ATCC No. 13637) possesses an exposed, immunologically accessible protein which binds to the Fc region of several species of immunoglobulins. Whole bacteria suspensions were incubated for 18 h with purified 125I-labelled antibodies with and without added non-labelled immunoglobulins. The suspensions were centrifuged for 30 rain and the pellet containing bacteria was assessed for radioactivity. Using this crude assay, the whole organism bound 125I-labelled rabbit and mouse immunoglobulins and the purified Fc portion of human IgG. All of these labelled preparations were competitively displaced by unlabelled rabbit and mouse immunoglobulins, and Fc of human IgG, as well as human immunoglobulin subclasses. The organism was sonicated to solubilize this immunoglobulin binding protein. Using this sonicated preparation, it was shown that unlabelled Fc of IgG, unlabelled mouse and rabbit immunoglobulins, all competitively displaced tzsI-labelled human Fc of IgG in a dose-response manner. A partially purified protein was prepared by Sephacryl S-300 followed by Sephadex G-100 column chromatography. This preparation was incubated with 125I-Fcy and with the following purified unlabelled preparations: F(ab') 2 of IgG, Fc of IgG, murine monoclonal lgA, IgG1, IgG2, IgG3, and IgG4. All except F(ab') 2 of IgG produced dose response competitive displacement. The molecular weight, as estimated by SDS-PAGE and Western blot, was 30,000 daltons. In Western blots, Fcy, murine monoclonal IgA, and human immunoglobulin subclasses, all showed affinity for the immobilized protein. Human F(ab') 2 fragments did not show affinity for the protein. Radioiodinated pseudomonal Ig-binding protein showed affinity for human IgG coupled to Sepharose, and was displaced by unlabelled pseudomonal Ig-binding protein. Scatchard analysis of binding showed two binding affinities: two distinct types of Ig-binding proteins were obtained, a high affinity with K a = 1.54 × 10-10 and a lower affinity with Kj = 2.36 x 10 -s. This immunoglobulin binding protein may be useful in immunoglobulin purification or identificationl Key words: Pseudomonas; Ig binding; Fc region
Introduction
Various strains of Staphylococcus aureus carry a type I receptor, protein A, which binds the Fc Correspondence to: W.D. Odell, Department of Medicine, University of Utah School of Medicine, 50 North Medical Drive, Salt Lake City, UT 84132, U.S.A. (Tel.: (801) 581-7459).
region of IgG (Kronvall et al., 1970; Hjelm et al., 1972; Enderson, 1979; Inganas et al., 1981; Langone, 1982). Certain strains of Groups A, C, and G of streptococci possess a type II receptor on their surface which also binds the Fc region of IgG (Kronvall, 1973). Kronvall et al. (1983) have also reported a receptor on groups C and G streptococci which also binds the F(ab') 2 region
188 of IgG. In addition to the classical immunoglobulin binding proteins, the existence of binding proteins which bind to regions outside the Fc region have also been documented. There exist bacterial cell-wall proteins which bind large fractions of polyclonal IgM (Kronvall et al., 1988), with affinity for IgG light chains (Bjorck, 1988), as well as binding of K and h light chains to group A streptococci (Bjorck et al., 1986). Our studies were initially directed at characterizing chorionic gonadotropin-like materials (CG) produced by some bacteria. We attempted to identify CG-like material in Pseudomonas maltophilia (ATCC No. 13637), which had previously been reported to contain immuno-accessible CG-like molecules (Acevedo et al., 1978; Robinson et al., 1987). However, when we used anti-CG monoclonal antibodies in an equilibrium type radioimmunoassay (RIA) a paradoxical increase in bound label hormone was observed when increasing doses of unlabelled hormone were added. In such equilibrium RIAs, one expects to observe decreasing bound labelled hormone with increasing doses of unlabelled hormone. This inconsistency was attributed to an Ig-binding protein on the surface of Pseudomonas maltophilia. This protein binds immunoglobulins, by their Fc regions, of various species. It differs from other known immunoglobulin binding proteins, protein A from Staphylococcus aureus, and protein G from streptococci, in that the former does not bind IgG3, while the latter does not bind IgA. In this paper, we describe the isolation, binding properties, and affinity of an Ig-binding protein from Pseudomonas
maltophilia.
Materials and methods
Bacterial cultures and preparation Multiple batches of four to six grams wet weight of Pseudomonas maltophilia (ATCC No. 13637) were grown on serum free GC agar (Difco) plus 2% v / v Isovitalex (BBL) at 37~C in 5% CO z, in air. The bacteria were scraped off the agar surface and weighed prior to use. Whole bacteria were suspended in 0.01 M phosphate, 0.15 M NaC1, pH 7.4 buffer (PBS) to achieve a bacterial concentration of 1.8 m g / 1 0 0 p.1. 100 /~1 were
pipetted into a series of 12 × 75 mm polypropylene tubes for binding studies.
Antibodies Several species and types of purified immunoglobulins were prepared in our laboratory and labelled with 125I employing a modification of the chloramine-T method (Hunter et al., 1962), as previously described from our laboratory (Odell et al., 1980). Immunoglobulins were radiolabelled and average specific activity of the radiolabelled antibodies were: (a) rabbit anti-CG (26 p.Ci/#g); (b) rabbit anti-goat immunoglobulin (30/zCi/p.g); (c) mouse monoclonal anti-CG (16 /zCi//.~g); (d) mouse monocl0nal anti-c~-CG (24 /zCi//.~g); and (e) Fc of human IgG (38 p.Ci//~g). Polyclonal rabbit anti CG and rabbit anti-goat immunoglobulin were purified by Staphylococcus-protein A chromatography (Fuller et al., 1987). The monoclonal mouse anti-a-CG and anti-CG were also purified by protein A chromatography. Fc of human IgG, F(ab') 2 fragments of human IgG and human IgG1, IgG2, IgG3, IgG4, were purchased from Chemicon (El Segundo, CA). The IgG1, IgG2, IgG3, and IgG4 are listed as > 95% pure. The Fc fragment is reported to give a single band by immunoelectrophoresis. However, to ascertain whether various human IgG subclasses could be contaminated with small amounts of human IgA, as these subclasses were purified from serum, we ran 12% polyacrylamide gels with these human immunoglobulin subclasses, and stained them with Coomassie blue R-250. A second gel was Western blotted onto nitrocellulose and hybridized with 12~I-mouse anti-human IgG (Fcspecific) antibodies purchased from Chemicon (El Segundo, CA) and also with 125I-mouse anti-human IgA antibodies. The blots were autoradiographed at - 7 0 ° C using a Kodak X ray film for 4 days. No contamination of human IgG subclasses with human IgA was observed. Myeloma cell lines TIB194/2F.11.15 producing murine monoclonal IgA antibody to trinitrophenol were purchased from the American Type Culture Collection (Rockville, MD). Cell lines were grown in RPMI medium 1640 with 10% fetal calf serum. The immunoglobulins were separated from fetal calf serum by protein A chromatography (Fuller et al., 1987). To ascertain
189 whether the calf serum is free of immunoglobulins, aliquots of protein A purified calf serum were run on a 15% polyacrylamide gels, and subsequently transblotted onto nitrocellulose. The membrane was hybridized with 3 × l0 s c p m / m l of ~25I-mouse anti-IgG antibody for 48 h. The membranes were washed to remove unbound radioactivity and exposed to a Kodak X ray film for 2 days. We found that after four subsequent protein A purification steps, the immunoglobulins present in calf serum were removed to a level not to be detected by the Western blot method. The monoclonal IgA produced by the myeloma cell lines was present in the media and was purified by protein A chromatography. The yield of purified IgA was variable, yielding only about 20-30% (Langone, 1982).
Proteins A and G Staphylococcus aureus-protein A was purchased from Sigma Chemical (St. Louis, MO) and streptococcal protein G was purchased from Chemicon (El Segundo, CA).
Whole bacteria binding studies 100 /zl containing 1.8 mg of whole bacteria (wet weight) were pipetted into a series of 12 x 75 mm polypropylene tubes. The 125I-immunoglobulin being studied was pipetted into the tubes, along with varying amounts of unlabelled immunoglobulins, dissolved in 0.01 M phosphate, 0.15 M sodium chloride pH 7.4 buffer (PBS). The buffer in this and subsequent studies contained 0.15% BSA as a carrier protein. This percentage was selected by optimizing the assay using different percentages of BSA. The presence of BSA decreased non-specific binding. Total volume in each tube was 1 ml. After 18 h incubation, the tubes were centrifuged at 4000 r.p.m, for 30 min. The supernatant was discarded and the pellet washed twice with PBS, 0.15% BSA. The radioactivity in the pellet was quantified in a gammaspectrometer. In some studies, the weights of the whole bacteria were varied.
Sonicated preparation of bacteria Approximately 4 g of a fresh batch of Ps. maltophilia were suspended in 6 vols. of PBS. This preparation was subjected to sonication in a
Bransonic sonicator for a period of 1 h a t room temperature after which the preparation was centrifuged at 4000 r.p.m, for 10 min and the supernatant retained for binding studies. The period of sonication was optimized for 1 h because sonication beyond a period of 60-80 min resulted in decrease and ultimately loss of activity of the protein.
Binding studies with the sonicated preparation The protein concentration of the supernatant fluid from the sonicate was determined by the method of Lowry et al. (1951). The sonicate was studied by incubation with various 125I-labelled and unlabelled immunoglobulins, such as Fc of IgG (10 txg), rabbit anti-CG (100 tzg), mouse anti-CG No. 9 (10 ~zg), mouse anti-a-hCG (10 Ixg), rabbit anti-goat IgG (10/xg). In another set of experiments the sonicate was incubated with unlabelled IgA (10 /xg), F(ab') 2 of IgG (10 /~g), IgG1 (5 /xg), IgG2 (5 /zg), IgG3 (6 tzg), IgG4 (5 ~g). Briefly, 1 ~g of the sonicate supernatant was pipetted into 12 × 75 mm polypropylene tubes. The radiolabelled immunoglobulin probe to be tested along with the respective unlabelled antibody was added to it. The final volume of 1 ml was obtained by adding PBS and 0.15% BSA. After an 18 h incubation at room temperature, the tubes were centrifuged at 10,000 × g for 30 min. The supernatant was discarded and the pellet washed once with PBS. The radioactivity of the pellet was determined by a gammaspectrometer.
Binding/kinetic studies using partially purified protein 50 mg of the sonicate supernate in 5 ml PBS were subjected to Sephacryl S-300 column chromatography, 45 cm × 2.5 cm with a flow rate of 55 m l / h , pre-equilibrated with PBS, and 3.5 ml fractions were collected. The protein concentrations of each fraction were estimated by measuring the optical density at 280 nm by a UV spectrophotometer. The immunoglobulin binding property was determined by incubating 1 ml of each fraction with radioiodinated Fc portion of human IgG, in 12 × 75 mm polypropylene tubes. After 18 h incubation at room temperature, the tubes were centrifuged at 10,000 × g for 30 min.
190 The supernatant was discarded and the pellets were counted for radioactivity in a gammaspectrometer with an efficiency of 70%. The tubes exhibiting immunoglobulin binding were pooled and lyophilized overnight. The lyophilate was resuspended in PBS, and its protein concentration was again determined. This protein mixture was then passed through a Sephadex G-100 column with PBS, of length 105 × 1 cm and a flow rate of 18.5 m l / h . 2 ml fractions were collected and analyzed for protein concentration by determining optical density at 280 nm by a UV spectrophotometer. The immunoglobulin binding property was determined as described earlier. The tubes containing activity were pooled and lyophilized. Using this partially purified preparation, binding/displacement studies were performed in a similar manner to that of the whole bacteria and sonicated proteins. The radiolabelled probe in this study was Fc of human IgG while displacement was studied by using highly purified unlabelled Fc of human IgG, F(ab') 2 of IgG, murine monoclonal IgA, and various immunoglobulin subclasses, i.e., IgG1, IgG2, IgG3, and IgG4. After incubation for 18 h, the tubes were centrifuged at 10,000 × g and the supernatant aspirated, cpm present in the pellet was assessed in a gammaspectrometer. To compare the specificity of immunoglobulin binding with known immunoglobulin binding proteins, the above studies were repeated by using staphylococcal protein A, streptococcal protein G, and as negative controls, two non-immunoglobulin binding proteins, bovine serum albumin (BSA) (Fraction V-RIA grade, Sigma Chemical, St. Louis, MO), and human transferrin (Sigma Chemical, St. Louis, MO) were also used. Bound protein was separated from non-bound ~25I-Fc of IgG by centrifugation at 10,000 × g for 30 min. The equilibrium constant of the Ig-binding protein was determined by the method used by Bjorck et al. (1986), but with a slight variation. The Pseudomonas Ig-binding protein was radioiodinated with 125I by the modified chloramine-T method described previously. The specific activity was about 16 ~ C i / ~ g . About 10 mg of human IgG was conjugated to 10 ml of Sepharose beads, yielding a concentration of 1
m g / m l of IgG:beads. Here 20,000 cpm of 1251Pseudomonas Ig-binding protein, 60 /~l of Sepharose-IgG beads, and different concentrations (0.01-100 /~g) of unlabelled Pseudomonas Ig-binding protein were mixed. This final volume of 0.5 ml was adjusted by adding PBS, 0.15% BSA. First a time study was performed to determine the hours required to achieve 50% and 100% binding. The assay tubes were incubated for 18 h at 37 ° C centrifuged, and the radioactivity in the beads counted. The bound and free concentrations were calculated. Results were plotted, and analyzed by the method described by Scatchard, (1949).
Page / Western blot studies 25 /~g of the sonicated preparation, and the partially pure protein were electrophoresed on a 12% polyacrylamide gel. Samples were run under denaturing conditions, containing both SDS and mercaptoethanol. In addition, samples were boiled for 10 minutes before application. Following electrophoresis, the gel was stained with Coomassie blue R-250. Samples containing the partially pure protein were electrophoresed on a 12% polyacrylamide gel and transblotted onto nitrocellulose membranes (Amersham) (Towbin et al., 1979). The membranes were hybridized for 24 h with the following radioiodinated probes (2 × 105 cpm/ml), Fc of IgG, F(ab') 2, murine monoclonal IgA, and various immunoglobulin subclasses, i.e., IgG1, IgG2, IgG3, and IgG4. After washing away the unbound radioactivity, the membranes were autoradiographed for 4-8 h at - 70 ° C by using a Kodak X ray film on an image intensifier.
Results
Binding studies with whole bacteria The bacteria bound ~25I-immunoglobulin from three species in this system (cpm bound increased as the amount of 125I-immunoglobulin was increased). The percent of labelled immunoglobulin bound increased as the amount of bacteria added was increased. The 125I-immunoglobulin was competitively displaced not only by the same unlabelled immunoglobulin as was labelled, but
191 10000
y////~
Y////,
Y////~
80OO
////// ////// //////
////// /i//// //////
6OOO o
/////I //////
o
4000
I///11 I/ // // /f /l /l I////I IIIIIi //11/l ~111.. IIIIII ////// ////// //////
2000-
~ *Rb.
U,/,/,/,. /x///x/ // x/ /
V/Ill,
/ / / / / / ~
I///I/ //////
ffZ777-~...... t111/i /////i
"Rb.+Rb.
*MA.
vzzg/( "MA+MA
*R
*R+R
"MCG "MCG+MC
*Fc
FC
1
*F(ab)2
22,500CPMOFANTIBODYPROBEADDED
Fig. 1. This graph shows the binding of various 12sI-immunoglobulin probes to 1.8 mg of Ps. maltophilia, and their competitive displacement with unlabelled immunoglobulins of various species. Note that 12SI-radiolabelled F(ab') 2 of IgG did not bind to the bacteria. (a) * Rb = 12SI-labelled polyclonal, rabbit anti-CG. (b) * Rb + Rb = x25I-labelled rabbit anti-CG in presence of 100 txg of unlabelled rabbit anti-CG. (c) * MA = 12SI-labelled monoclonal, mouse anti-c~-CG. (d) * MA + MA = 125I-labelled mouse anti-a-CG in presence of 10/xg of unlabelled mouse anti-a-CG. (e) * R = tzsI-labelled polyclonal, rabbit anti-goat IgG. (f) * R + R = tzSI-rabbit anti-goat IgG in presence of 10/xg of unlabelled rabbit anti-goat IgG. (g) * MCG = t2SI-labelled mouse anti-CG. (h) * MCG + MCG = lzSI-labelled monoclonal, mouse anti-CG in presence of 10/xg of unlabelled mouse anti-CG. (i) * Fc = 12sI-labelled Fc of IgG. (j) Fc = 12SI-labelled Fc of IgG in presence of 10/xg of unlabelled Fc of IgG. (k) * F ( a b ' ) 2 = 12SI-labelled F(ab') 2 of IgG.
also by unlabelled immunoglobulin from a different species. Thus, in Fig. 1 are shown the binding and competitive displacement of labelled rabbit, and mouse antibodies and of labelled Fc of IgG. This binding and displacement did not relate to the antigenic determinants recognized by the immunoglobulins, but only to the fact that they were immunoglobulins. Thus, both rabbit anti-CG and rabbit anti-goat immunoglobulin bound and were competitively displaced. Similarly, both mouse monoclonal anti-CG and anti-a-glycoprotein subunit bound and were competitively displaced. The Fc portion of human IgG also bound and was competitively displaced. Unlabelled F(ab') 2 fragments did not show binding.
Immunoglobulin dose-response curves with whole bacteria The next set of studies was performed by incubating 1.8 mg of whole bacteria with the five ~2SI-radiolabelled antibodies, in the presence of increasing amounts of the respective unlabelled immunoglobulin. The percent labelled immunoglobulin increased from an average of 8 15% when the amount of bacteria was doubled from 0.9 to 1.8 mg per tube. For all five antibod-
ies (two mouse, two rabbit and the Fc portion of human IgG), there was a dose-response related competitive displacement of the 12SI-labelled immunoglobulin or Fc fragment (data not shown). The amount of immunoglobulin producing maximal displacement of label varied as follows: human F c y = 1.8 /zg, rabbit anti-CG = 1 ~g, rabbit anti-goat IgG = 10 ~g, mouse anti-CG = 10/xg.
Binding studies using partially purified Fc-binding protein The sonicate supernate prepared as described in the materials and methods section was partially purified by serial Sephacryl S-300 followed by Sephadex G-100 column chromatography. The immunoglobulin binding property of the eluted fractions of both the Sephacryl S-300 and Sephadex G-100 showed that it was dispersed over a range of 85 kDa to 60 kDa (data not shown). We observed that there was an aggregation of the pure protein, after the labelled immunoglobulin was added to it, thus, protein is forming a much larger complex with the 125I-Fc fragment, making it large enough to be centrifuged at 10,000 x g, the step used to separate bound and free 125I-Fcy.
192 MYELOMA Fc/F(ab)2
120
120"
•
A---O-.dk ~--
•
[]
Prot.A.(Fc)
•
¢--.__~ el ', a ~ . . _ . , ~ D
100
IgA
0
F(=b)2
o
Prot.G.(Fc)
A
P.Malt.(Fc)
[]
ProtJ,.(F(ab))
•
Prot.G.(F(ab))
•
P,Malto.(F(ab))
100
0 o
~,A
°o
o
•
Prot. A,
o
Prot. G.
A
P. Malto.
•
Prot. A.
o
Prot. G.
&
P. Malto.
\\ 80
80
o<
'ii,
O ¢D
40
o-e o co
:Z
so
60-
co
',x,
.
40,
',\ \11"~ "\"A
ix
20
oN
o~
A 0
B
" ~ O~o
........
.01
FC
"R
\ ,
........
,
.1
.......
1
~
10
.......
,
100
.0001
.001
.01
.1
1
10
DOSE OF IMMUNOGLOBULIN (~lg)
DOSE OF IMMUNOGLOBULINFRAGMENT(pg)
IgG2 IgG1 120"
120
100
100
\ \
,,
80
•
Prot.A.
o
Prot.G.
A
P. Malto.
\\,.
80"
i~
o-e O
m
60
40 l'i
60o CD
40-
20'
20
C •0001
"~~---8.001
.01
.1
1
10
DOSE OF IMMUNOGLOBULIN FRAGMENT(pg)
.0001 .001
.01
.1
1
10
100
DOSE O F I M M U N O G L O B U L I N FRAGMENT(~Jg)
Fig. 2 (continued on next page)
193
IgG4
IgG3 120"
120 O A,
• •
•
100
100
Prot.A.
~o.~
•
\
o Prot.G. 01
80
o~
A
l
I 1
& P. Malto.
',
Prot.A.
o Prot.G. P. Malto.
so. I/
'1
o
0
60"
m
40"
!I
I
\\, 2o.
"~
"..o~_,
E
.001
.01
.1
1
10
DOSE OF IMMUNOGLOBULIN FRAGMENT(pg)
o
...................................
0"s 1 0"4 1 0"3 10"2 10"1 1 0 ° 101
DOSE OF IMMUNOGLOBULIN FRAGMENTgJg )
Fig. 2. These graphs show the binding and competitive displacement of 125I-labelled Fc of IgG by increasing doses of unlabelled Fc, myeloma IgA and various immunoglobulin subclasses on incubation with 1 /~g of the partially purified Pseudomonas protein. The cpm bound are all normalized to 100%, defined as cpm bound in tubes containing no added unlabelled immunoglobulin or immunoglobulin fragments. All the graphs show a dose-response relationship except for F(ab') 2 which does not compete for binding of 125I-Fc. Staphylococcal protein A and streptococcal protein G are run alongside for each immunoglobulin understudy as a comparison. A: dose-response curve showing binding of 125I-Fc7 with its competitive displacement by unlabelled Fc3, and F(ab') 2 with the Pseudomonas partially pure protein, protein A and protein G. B: dose-response curve showing binding of 125I-Fc-/ with its competitive displacement by increasing doses of unlabelled murine monoclonal IgA. C: dose-response curve showing binding of 12SI-Fcy with its competitive displacement by unlabelled human IgG1. D: dose-response curve showing binding of 12Sl-Fc7 with its competitive displacement by increasing doses of unlabelled human IgG2. E: dose-response curve showing binding of 125I-Fc7 with its competitive displacement by increasing doses of unlabelled human IgG3. F: dose-response curve showing binding of ~251-Fc7 with its competitive displacement by increasing doses of unlabelled human IgG4.
Dose-response relationships were observed when the partially pure protein was incubated with 125I-Fc of human IgG, and increasing doses of unlabelled Fc of human IgG, (F(ab') 2 of human IgG, murine monoclonal IgA, and human IgG1, ]gG2, IgG3, IgG4. All the immunoglobulin fractions displaced radiolabelled Fc except F(ab') 2 of IgG. These data are shown in Figs. 2A-2F. Similar dose-response relationships were also observed when protein A of Staphylococcus or protein G of Streptococcus was used as binding proteins. For all the above studies the concentra-
tions of protein A, protein G, and the Pseudomonas protein were normalized by first performing titres to give 25-27% binding of tzsI-Fc. With two exceptions, all immunoglobulins competed for tzSI-Fc binding in identical dose response relations for all three immunoglobulin binding proteins. For protein G, murine monoclonal IgA failed to compete for Fc binding, while for Pseudomonas protein, IgA inhibited Fc binding. For staphylococcal protein A, IgG3 failed to compete for Fc binding, while for Pseudomonas protein, IgG3 inhibited Fc binding. Unlabelled human F(ab') 2 did not compete for label binding.
194 100.
TABLE I EQUILIBRIUM CONSTANTS OF VARIOUS PROTEINS U S I N G H U M A N IgG
8,.I
Equilibrium constant ( h u m a n IgG) Protein A a Protein G a Pseudomonal b Ig-binding protein
6.7× 10 -1° 4.4× 10 -1° 1.54 × 1 0 - 1 ° / 2 . 3 6 × 10-8 c
a Akerstrom and Bjorck, 1986. Determined by Scatchard analysis using h u m a n IgG coupled to Sepharose beads and radioiodinated p s e u d o m o n a s Ig-binding protein. c Two slopes were obtained in the Scatchard plot.
The control proteins, BSA and transferrin, showed no binding of labelled immunoglobulin (data not shown). The kinetics of I g G - P s e u d o m o n a s Ig-binding protein was used using human I g G coupled to Sepharose. Our initial time studies showed that the binding is slow, reaching 50% after nearly 10 h. From here the binding was even slower, saturating at 18-20 h. This interaction was studied at 4 ° C and 37 ° C, using h u m a n IgG. The binding at 4 ° C greatly decreased, taking nearly 18 h to achieve 50%. Table I compares the equilibrium constants of the Pseudomonas Ig-binding protein with that of protein A and protein G. As the equilibrium constant ( K d) of the reaction is equal to the absolute value of the slope of the curve, we identified two distinct types of Ig-binding proteins which are present in Pseudomonas. One has a very high affinity ( K d = 1.545 × 10-10) and one with a low affinity ( K d = 2.36 × 10-8). As seen in Table I, one type of the p s e u d o m o n a s Ig-binding protein has an equilibrium constant in the same magnitude as that of protein A and protein G. There also exists a binding protein on Pseudomonas of very low affinity. The p r e p o n d e r a n c e of which may explain the length of time (10 h) required to achieve 50% saturation as c o m p a r e d to protein A, protein G. Fig. 3 shows the Scatchard diagram of the Ig-binding protein, using human I g G coupled to Sepharose and radiolabelling the Pseudomonas Ig-binding protein. H e r e two slopes were obtained: one having a high affinity ( K d = 1.545 x 10 - l ° ) and one with a low affinity (K~ = 2.36 × 10-8).
.545~10-10 6o
I1 m 4o,
1I 1
""" 19
39
59
79
99
119139159179199
BOUND (P. MOL.) Fig. 3. The figure shows the Scatchard plot for the reaction between the pseudomonas Ig-binding protein and human IgG coupled to Sepharose beads. Here the equilibrium constant (K d) of the reaction is equal to the absolute value of the slope of the curve. Two different types of Ig-binding protein were seen: one with high affinity (K d = 1.545 × 10-10) and one with low affinity (K d = 2.36× 10-8).
PAGE / Western blot studies Fig. 4A shows a 12% polyacrylamide gel stained with Coomassie blue. Lane 1 shows the profile of the sonicated proteins. Lane 2 shows the profile of the purified preparation. Note that there is a major band at 30 kDa and there are also minor bands at 69 and 5 kDa. Fig. 4B shows Western blot analysis of the purified protein hybridized with various radiolabelled probes. As seen in Fig. 4B, F c y murine monoclonal IgA, various h u m a n i m m u n o g l o b u l i n subclasses showed reaction towards the immobilized protein, whereas human F(ab') 2 did not show any reaction. The major band identified in Fig. 4A, lane 2, was identified to be the immunoactive part at 30 kDa.
Discussion Proteins, with affinity for different regions of mammalian IgG, have been isolated and well characterized from different strains of bacteria. The ability of Protein A, derived from Staphylo-
195
MOL WT Kd
I
2
A)
93B) MOL WT
69-
Kd
I
2
3
4
5
6
7
4593--
3069-45--
2130--
14.3-
21~
14.3--
_
5--
Fig. 4. A: 25 /.~g of the sonicated and purified material were electrophoresed on a 12% polyacrylamide gel, and stained with Coomassie blue R-250. Lane 1 shows the profile of the sonicated protein with a wide range of molecular weights, whereas, after purification there is a large prominent band at 30 kDa region with minor bands at 69 kDa and 5 kDa regions. B: the figure shows the Western blot analysis of the purified Ig-binding material, hybridized for 24 h, with the following radiolabelled human immunoglobulin fragments and subclasses. Lane 1, x25I-Fcy; lane 2, 125I-murine monoclonal IgA; lane 3, 125I-IgG1; lane 4, 125I-IGG2; lane 5, 125I-IGG3; lane 6, 125I-IGG4; lane 7, 1251-F(ab')2.
coccal aureus, to bind IgG via its Fc region has made it an important immunological tool. In addition, streptococci groups A and C possess a surface protein which binds the Fc regions of IgG (Kronvall, 1973), and groups C and G have surface proteins which bind the F(ab') 2 regions of IgG (Kronvall et al., 1983). Very recently Lindahl and Akerstrom (1989) have described the existence of an IgA binding protein in group A Streptococcus. Strains of Streptococcus and Peptococcus magnus have also been identified which possess surface proteins, which preferentially bind light chains of immunoglobulins (Bjorck et al., 1986; Bjorck, 1988). We have also demonstrated that Ps. maltophilia also has an immunoaccessible surface protein which binds IgG via their Fc regions, murine monoclonal IgA, but not F(ab') 2 of IgG. As described in the materials and methods section, we could not detect any contamination of the human IgG subclasses with small amounts of human IgA by the Western blot method. We also eliminated the possibility of serum IgA anti-pseudomonal antibodies giving
false binding in our studies by using IgA secreted by an ATCC myeloma cell line which secrets only monoclonal IgA antibodies directed against trinitrophenol. Thus, the binding of IgA by the Pseudomonas protein appears to be a valid observation. We observed a discrepancy between the molecular weight of this protein as estimated by Western blot analysis which gave a weight of 30 kDa, and by the elution profile of the immunoglobulin binding property of Sephacryl S-300 and Sephadex G-100 chromatography, which gave a weight range of 60-80 kDa. Here the elution profile was polydispersed over a wide range of tubes rather than a sharp peak (data not shown). This could be possible if the protein molecule is aggregating into large polymeric complexes or that it is associated with a heavier protein. The gel electrophoresis is performed under denaturing conditions, thus, the aggregated protein could be denatured into its individual components, one of which is immunoactive. We tested for presence of Fc binding activity in the supernate of the partially pure protein after centrifugation at
196
10,000 x g and could detect none. This finding would also suggest that the native protein may exist in an aggregated form or associated with a larger MW protein. We observed large insoluble complexes with polyclonal IgG. This would indicate that the Ig-binding protein has several binding sites akin to protein G which has at least 2/molecule and protein A, 4/molecule. Scatchard analysis showed that there are two types of the protein, one with high affinity (K d = 1.545 × 10 -1°) and one with low affinity (K d = 2.36× 10 8), when using human IgG. The low affinity may exist in much larger quantity, thus explaining the 20 h period of incubation required to attain saturation of binding to IgG-Sepharose. This Pseudomonas protein differs in the spectrum of immunoglobulins bound from that of staphylococcal protein A in that it binds human IgG3, while protein A does not bind IgG3. The pseudomonal immunoglobulin binding protein also differs from protein G in that the former binds IgA while protein G does not bind IgA. These properties suggest the three binding proteins are products of different genes. We discovered the Fc-binding protein in the course of studies designed to evaluate reports that a CG-like protein was produced by Pseudomonas. Acevedo et al. (1978) have reported this observation based on anti-CG antibody binding to Pseudomonas. Richert and Ryan (1977) have reported that CG-like receptors exist on the surface of Ps. mahophilia, a finding that does suggest CG-like materials may effect Pseudomonas actions, growth or some other function. In lieu of the presence of immunoglobulin binding proteins on the surface of Ps. maltophilia, it is difficult to interpret antibody binding identification of CG in Pseudomonas. We found considerable artifactual effects when CG immunoassays were used to study possible CG production by Ps. maltophilia. The current studies indicate that the presence of any surface molecule on bacteria demonstrated by binding of free antibodies does not prove the presence of the molecule unless the studies also demonstrate that the antibody binding is specific and is not displaced by Fc fragments. Further, the current studies demonstrate an additional desirable feature of the use of gold
immunoprobes to explore bacterial surfaces. Gold immunoprobes, as conventionally constructed (Robinson et al., 1987), have the Fc portion oriented toward the center of the probe and display only or primarily the F(ab') 2 portions of bound antibodies. Thus, false positive binding of antibodies by their Fc portions (as demonstrated with free antibodies in the current paper) is sterically hindered. The wider spectrum of immunoglobulin classes bound by the Ps. mahophilia relative to staphylococcal protein A may make this protein a useful reagent in immunoabsorbent column chromatography and gold or silver immunoprobe construction.
Note added in proof
Pseudomonas maltophilia was transferred to the genus Xanthomonas as Xanthomonas maltophilia (Hugh 1981) (Swings, J., De Vos, P., Van der Nooter, M. and De Ley, J. (1983) Int. J. Syst. Bacteriol. 33, 409).
Acknowledgements We would like to thank Gary Gorby, Jeanine Griffin, and John Klein for their technical assistance and cooperation. We thank the late Virgie Roubidoux and Char Kidd for editorial assistance and preparation of this manuscript. This paper is based upon work supported by National Institutes of Health Grant no. RO1 HD18986.
References Acevedo, H.F., Slifkin, M., Pouchet, G.R. and Pardo, M. (1978) Immunohistochemical localization of choriogonadotropin-like protein in bacteria isolated from cancer patients. Cancer 41, 1217. Bjorck, L. (1988) Protein L, A novel bacterial cell-wall protein with affinity for Ig L chains. J. Immunol. 140, 1194. Bjorck, L. and Akerstrom, B. (1986) A physiochemical study of Protein G, a molecule with unique immunoglobulin G-binding properties. J. Biol. Chem. 261, 10240. Bjorck, L., Logdberg, L., Berggard, B., Schalen, C. and Persson, M.H. (1986) Related bindings of aggregated /32-mi-
197 croglobulin, IgG Fab, kappa and lambda light chains to group A Streptococci. Acta Pathol. Microbiol. Immunol. Scand. 94, 25. Enderson, C. (1979) The binding to Protein A of immunoglobulin G and of Fab and Fc fragments. Acta. Pathol. Microbiol. Scand. 87, 185. Fuller, S.A., Takahashi, M. and Hurrell, J.G.R. (1987) Purification of monoclonal antibodies. In: F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith and H. Struhl (Eds.). Current Protocols in Molecular Biology, Vol. II. Wiley, New York, pp. 11.11.1-5. Hjelm, H., Meloun, B. and Sjoquist, J. (1972) Protein A isolated from Staphylococcal aureus after digestion with lysostaphin. Eur. J. Biochem. 29, 572. Hunter, W.M. and Greenwood, F.C. (1962) Preparation of iodine 131 labelled human growth hormone of high specific activity. Nature 194, 495. Inganas, M. and Johansson, S.G.O. (1981) Influence of the alternative protein A interaction on the precipitation between human monoclonal immunoglobulins and protein A from Staphylococcus aureau. Int. Arch. Allergy Appl. Immunol. 65, 91. Kronvall, G. (1973) A surface component in group A, C, and G Streptococci with non-immune reactivity for immunoglobulin G. J. Immunol. 111, 1401. Kronvall, G. and Gunnar, L. (1988) Non-immune binding of Ig to clostridium perfringeus, preferential binding of IgM and aggregated IgG. J. Immunol. 140, 1223. Kronvall, G., Paul, G.Q. and Williams, Jr., R.C. (1970) Quantitation of Staphylococcal Protein A: determination of equilibrium constant and number of Protein A residues on bacteria. J. Immunol. 104, 273.
Kronvall, G., Myhre, E.B. and Erntell, M. (1983) Alternative non-immune F(abI)2 mediated immunoglobulin binding to group C and G Streptococci. Scand. J. Immunol. 17, 201. Langone, J.J. (1982) Protein A of Staphylococcus aureus and related immunoglobulin receptors produced by streptococci and pneumococci. Adv. Immunol. 32, 157. Lindahl, G. and Akerstrom, B. (1989) Receptor for IgA in Group A Streptococci: cloning of the gene and characterization of the protein expressed in Escherichia coli. Mol. Microbiol. 3, 239. Lowry, O.H., Rosebrough, N.J. Farr, A.L. and Randall, R.J. (1951) Protein measurement with the folin phenol reagent. J. Biol. Chem. 193, 265. Odell, W.D. and Heber, D. (1980) Radioiodination of peptides and proteins for radioimmunoassays or radioreceptor assays. In: N.R. Rose and H. Friedman (Eds.), Manual of Clinical Immunology, 2nd edn. American Society for Microbiology, Washington, DC, p. 329. Richert, N.D. and Ryan, R.J. (1977) Specific gonadotropin binding to Pseudomonas maltophilia. Proc. Natl. Acad. Sci. U.S.A. 74, 878. Robinson, Jr., E.N., McGee, Z.A. and Clemens, C.M. (1987) Probing the surface of bacteria: use of gold sphere immunological probes. Microbiol. Pathogen. 2, 159. Scatchard, G. (1949) The attractions of proteins for small molecules and ions. Ann. N.Y. Acad. Sci. U.S.A. 51, 660. Towbin, H., Staehlin, T. and Gordon, J. (1979) Electrophorectic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. U.S.A. 76, 4350.
187
JIM06010
Isolation of a 30 kDa immunoglobulin binding protein from Pseudomonas maltophilia Sanjeev Grover, Zell A. M c G e e and William D. Odell Departments of lnternal Medicine and Physiology, University of Utah School of Medicine, Salt Lake City, UT 84132, U.S.A. (Received 12 November 1990, revised received 28 February 1991, accepted 11 April 1991)
We have demonstrated that Pseudomonas maltophifia (ATCC No. 13637) possesses an exposed, immunologically accessible protein which binds to the Fc region of several species of immunoglobulins. Whole bacteria suspensions were incubated for 18 h with purified 125I-labelled antibodies with and without added non-labelled immunoglobulins. The suspensions were centrifuged for 30 rain and the pellet containing bacteria was assessed for radioactivity. Using this crude assay, the whole organism bound 125I-labelled rabbit and mouse immunoglobulins and the purified Fc portion of human IgG. All of these labelled preparations were competitively displaced by unlabelled rabbit and mouse immunoglobulins, and Fc of human IgG, as well as human immunoglobulin subclasses. The organism was sonicated to solubilize this immunoglobulin binding protein. Using this sonicated preparation, it was shown that unlabelled Fc of IgG, unlabelled mouse and rabbit immunoglobulins, all competitively displaced tzsI-labelled human Fc of IgG in a dose-response manner. A partially purified protein was prepared by Sephacryl S-300 followed by Sephadex G-100 column chromatography. This preparation was incubated with 125I-Fcy and with the following purified unlabelled preparations: F(ab') 2 of IgG, Fc of IgG, murine monoclonal lgA, IgG1, IgG2, IgG3, and IgG4. All except F(ab') 2 of IgG produced dose response competitive displacement. The molecular weight, as estimated by SDS-PAGE and Western blot, was 30,000 daltons. In Western blots, Fcy, murine monoclonal IgA, and human immunoglobulin subclasses, all showed affinity for the immobilized protein. Human F(ab') 2 fragments did not show affinity for the protein. Radioiodinated pseudomonal Ig-binding protein showed affinity for human IgG coupled to Sepharose, and was displaced by unlabelled pseudomonal Ig-binding protein. Scatchard analysis of binding showed two binding affinities: two distinct types of Ig-binding proteins were obtained, a high affinity with K a = 1.54 × 10-10 and a lower affinity with Kj = 2.36 x 10 -s. This immunoglobulin binding protein may be useful in immunoglobulin purification or identificationl Key words: Pseudomonas; Ig binding; Fc region
Introduction
Various strains of Staphylococcus aureus carry a type I receptor, protein A, which binds the Fc Correspondence to: W.D. Odell, Department of Medicine, University of Utah School of Medicine, 50 North Medical Drive, Salt Lake City, UT 84132, U.S.A. (Tel.: (801) 581-7459).
region of IgG (Kronvall et al., 1970; Hjelm et al., 1972; Enderson, 1979; Inganas et al., 1981; Langone, 1982). Certain strains of Groups A, C, and G of streptococci possess a type II receptor on their surface which also binds the Fc region of IgG (Kronvall, 1973). Kronvall et al. (1983) have also reported a receptor on groups C and G streptococci which also binds the F(ab') 2 region
188 of IgG. In addition to the classical immunoglobulin binding proteins, the existence of binding proteins which bind to regions outside the Fc region have also been documented. There exist bacterial cell-wall proteins which bind large fractions of polyclonal IgM (Kronvall et al., 1988), with affinity for IgG light chains (Bjorck, 1988), as well as binding of K and h light chains to group A streptococci (Bjorck et al., 1986). Our studies were initially directed at characterizing chorionic gonadotropin-like materials (CG) produced by some bacteria. We attempted to identify CG-like material in Pseudomonas maltophilia (ATCC No. 13637), which had previously been reported to contain immuno-accessible CG-like molecules (Acevedo et al., 1978; Robinson et al., 1987). However, when we used anti-CG monoclonal antibodies in an equilibrium type radioimmunoassay (RIA) a paradoxical increase in bound label hormone was observed when increasing doses of unlabelled hormone were added. In such equilibrium RIAs, one expects to observe decreasing bound labelled hormone with increasing doses of unlabelled hormone. This inconsistency was attributed to an Ig-binding protein on the surface of Pseudomonas maltophilia. This protein binds immunoglobulins, by their Fc regions, of various species. It differs from other known immunoglobulin binding proteins, protein A from Staphylococcus aureus, and protein G from streptococci, in that the former does not bind IgG3, while the latter does not bind IgA. In this paper, we describe the isolation, binding properties, and affinity of an Ig-binding protein from Pseudomonas
maltophilia.
Materials and methods
Bacterial cultures and preparation Multiple batches of four to six grams wet weight of Pseudomonas maltophilia (ATCC No. 13637) were grown on serum free GC agar (Difco) plus 2% v / v Isovitalex (BBL) at 37~C in 5% CO z, in air. The bacteria were scraped off the agar surface and weighed prior to use. Whole bacteria were suspended in 0.01 M phosphate, 0.15 M NaC1, pH 7.4 buffer (PBS) to achieve a bacterial concentration of 1.8 m g / 1 0 0 p.1. 100 /~1 were
pipetted into a series of 12 × 75 mm polypropylene tubes for binding studies.
Antibodies Several species and types of purified immunoglobulins were prepared in our laboratory and labelled with 125I employing a modification of the chloramine-T method (Hunter et al., 1962), as previously described from our laboratory (Odell et al., 1980). Immunoglobulins were radiolabelled and average specific activity of the radiolabelled antibodies were: (a) rabbit anti-CG (26 p.Ci/#g); (b) rabbit anti-goat immunoglobulin (30/zCi/p.g); (c) mouse monoclonal anti-CG (16 /zCi//.~g); (d) mouse monocl0nal anti-c~-CG (24 /zCi//.~g); and (e) Fc of human IgG (38 p.Ci//~g). Polyclonal rabbit anti CG and rabbit anti-goat immunoglobulin were purified by Staphylococcus-protein A chromatography (Fuller et al., 1987). The monoclonal mouse anti-a-CG and anti-CG were also purified by protein A chromatography. Fc of human IgG, F(ab') 2 fragments of human IgG and human IgG1, IgG2, IgG3, IgG4, were purchased from Chemicon (El Segundo, CA). The IgG1, IgG2, IgG3, and IgG4 are listed as > 95% pure. The Fc fragment is reported to give a single band by immunoelectrophoresis. However, to ascertain whether various human IgG subclasses could be contaminated with small amounts of human IgA, as these subclasses were purified from serum, we ran 12% polyacrylamide gels with these human immunoglobulin subclasses, and stained them with Coomassie blue R-250. A second gel was Western blotted onto nitrocellulose and hybridized with 12~I-mouse anti-human IgG (Fcspecific) antibodies purchased from Chemicon (El Segundo, CA) and also with 125I-mouse anti-human IgA antibodies. The blots were autoradiographed at - 7 0 ° C using a Kodak X ray film for 4 days. No contamination of human IgG subclasses with human IgA was observed. Myeloma cell lines TIB194/2F.11.15 producing murine monoclonal IgA antibody to trinitrophenol were purchased from the American Type Culture Collection (Rockville, MD). Cell lines were grown in RPMI medium 1640 with 10% fetal calf serum. The immunoglobulins were separated from fetal calf serum by protein A chromatography (Fuller et al., 1987). To ascertain
189 whether the calf serum is free of immunoglobulins, aliquots of protein A purified calf serum were run on a 15% polyacrylamide gels, and subsequently transblotted onto nitrocellulose. The membrane was hybridized with 3 × l0 s c p m / m l of ~25I-mouse anti-IgG antibody for 48 h. The membranes were washed to remove unbound radioactivity and exposed to a Kodak X ray film for 2 days. We found that after four subsequent protein A purification steps, the immunoglobulins present in calf serum were removed to a level not to be detected by the Western blot method. The monoclonal IgA produced by the myeloma cell lines was present in the media and was purified by protein A chromatography. The yield of purified IgA was variable, yielding only about 20-30% (Langone, 1982).
Proteins A and G Staphylococcus aureus-protein A was purchased from Sigma Chemical (St. Louis, MO) and streptococcal protein G was purchased from Chemicon (El Segundo, CA).
Whole bacteria binding studies 100 /zl containing 1.8 mg of whole bacteria (wet weight) were pipetted into a series of 12 x 75 mm polypropylene tubes. The 125I-immunoglobulin being studied was pipetted into the tubes, along with varying amounts of unlabelled immunoglobulins, dissolved in 0.01 M phosphate, 0.15 M sodium chloride pH 7.4 buffer (PBS). The buffer in this and subsequent studies contained 0.15% BSA as a carrier protein. This percentage was selected by optimizing the assay using different percentages of BSA. The presence of BSA decreased non-specific binding. Total volume in each tube was 1 ml. After 18 h incubation, the tubes were centrifuged at 4000 r.p.m, for 30 min. The supernatant was discarded and the pellet washed twice with PBS, 0.15% BSA. The radioactivity in the pellet was quantified in a gammaspectrometer. In some studies, the weights of the whole bacteria were varied.
Sonicated preparation of bacteria Approximately 4 g of a fresh batch of Ps. maltophilia were suspended in 6 vols. of PBS. This preparation was subjected to sonication in a
Bransonic sonicator for a period of 1 h a t room temperature after which the preparation was centrifuged at 4000 r.p.m, for 10 min and the supernatant retained for binding studies. The period of sonication was optimized for 1 h because sonication beyond a period of 60-80 min resulted in decrease and ultimately loss of activity of the protein.
Binding studies with the sonicated preparation The protein concentration of the supernatant fluid from the sonicate was determined by the method of Lowry et al. (1951). The sonicate was studied by incubation with various 125I-labelled and unlabelled immunoglobulins, such as Fc of IgG (10 txg), rabbit anti-CG (100 tzg), mouse anti-CG No. 9 (10 ~zg), mouse anti-a-hCG (10 Ixg), rabbit anti-goat IgG (10/xg). In another set of experiments the sonicate was incubated with unlabelled IgA (10 /xg), F(ab') 2 of IgG (10 /~g), IgG1 (5 /xg), IgG2 (5 /zg), IgG3 (6 tzg), IgG4 (5 ~g). Briefly, 1 ~g of the sonicate supernatant was pipetted into 12 × 75 mm polypropylene tubes. The radiolabelled immunoglobulin probe to be tested along with the respective unlabelled antibody was added to it. The final volume of 1 ml was obtained by adding PBS and 0.15% BSA. After an 18 h incubation at room temperature, the tubes were centrifuged at 10,000 × g for 30 min. The supernatant was discarded and the pellet washed once with PBS. The radioactivity of the pellet was determined by a gammaspectrometer.
Binding/kinetic studies using partially purified protein 50 mg of the sonicate supernate in 5 ml PBS were subjected to Sephacryl S-300 column chromatography, 45 cm × 2.5 cm with a flow rate of 55 m l / h , pre-equilibrated with PBS, and 3.5 ml fractions were collected. The protein concentrations of each fraction were estimated by measuring the optical density at 280 nm by a UV spectrophotometer. The immunoglobulin binding property was determined by incubating 1 ml of each fraction with radioiodinated Fc portion of human IgG, in 12 × 75 mm polypropylene tubes. After 18 h incubation at room temperature, the tubes were centrifuged at 10,000 × g for 30 min.
190 The supernatant was discarded and the pellets were counted for radioactivity in a gammaspectrometer with an efficiency of 70%. The tubes exhibiting immunoglobulin binding were pooled and lyophilized overnight. The lyophilate was resuspended in PBS, and its protein concentration was again determined. This protein mixture was then passed through a Sephadex G-100 column with PBS, of length 105 × 1 cm and a flow rate of 18.5 m l / h . 2 ml fractions were collected and analyzed for protein concentration by determining optical density at 280 nm by a UV spectrophotometer. The immunoglobulin binding property was determined as described earlier. The tubes containing activity were pooled and lyophilized. Using this partially purified preparation, binding/displacement studies were performed in a similar manner to that of the whole bacteria and sonicated proteins. The radiolabelled probe in this study was Fc of human IgG while displacement was studied by using highly purified unlabelled Fc of human IgG, F(ab') 2 of IgG, murine monoclonal IgA, and various immunoglobulin subclasses, i.e., IgG1, IgG2, IgG3, and IgG4. After incubation for 18 h, the tubes were centrifuged at 10,000 × g and the supernatant aspirated, cpm present in the pellet was assessed in a gammaspectrometer. To compare the specificity of immunoglobulin binding with known immunoglobulin binding proteins, the above studies were repeated by using staphylococcal protein A, streptococcal protein G, and as negative controls, two non-immunoglobulin binding proteins, bovine serum albumin (BSA) (Fraction V-RIA grade, Sigma Chemical, St. Louis, MO), and human transferrin (Sigma Chemical, St. Louis, MO) were also used. Bound protein was separated from non-bound ~25I-Fc of IgG by centrifugation at 10,000 × g for 30 min. The equilibrium constant of the Ig-binding protein was determined by the method used by Bjorck et al. (1986), but with a slight variation. The Pseudomonas Ig-binding protein was radioiodinated with 125I by the modified chloramine-T method described previously. The specific activity was about 16 ~ C i / ~ g . About 10 mg of human IgG was conjugated to 10 ml of Sepharose beads, yielding a concentration of 1
m g / m l of IgG:beads. Here 20,000 cpm of 1251Pseudomonas Ig-binding protein, 60 /~l of Sepharose-IgG beads, and different concentrations (0.01-100 /~g) of unlabelled Pseudomonas Ig-binding protein were mixed. This final volume of 0.5 ml was adjusted by adding PBS, 0.15% BSA. First a time study was performed to determine the hours required to achieve 50% and 100% binding. The assay tubes were incubated for 18 h at 37 ° C centrifuged, and the radioactivity in the beads counted. The bound and free concentrations were calculated. Results were plotted, and analyzed by the method described by Scatchard, (1949).
Page / Western blot studies 25 /~g of the sonicated preparation, and the partially pure protein were electrophoresed on a 12% polyacrylamide gel. Samples were run under denaturing conditions, containing both SDS and mercaptoethanol. In addition, samples were boiled for 10 minutes before application. Following electrophoresis, the gel was stained with Coomassie blue R-250. Samples containing the partially pure protein were electrophoresed on a 12% polyacrylamide gel and transblotted onto nitrocellulose membranes (Amersham) (Towbin et al., 1979). The membranes were hybridized for 24 h with the following radioiodinated probes (2 × 105 cpm/ml), Fc of IgG, F(ab') 2, murine monoclonal IgA, and various immunoglobulin subclasses, i.e., IgG1, IgG2, IgG3, and IgG4. After washing away the unbound radioactivity, the membranes were autoradiographed for 4-8 h at - 70 ° C by using a Kodak X ray film on an image intensifier.
Results
Binding studies with whole bacteria The bacteria bound ~25I-immunoglobulin from three species in this system (cpm bound increased as the amount of 125I-immunoglobulin was increased). The percent of labelled immunoglobulin bound increased as the amount of bacteria added was increased. The 125I-immunoglobulin was competitively displaced not only by the same unlabelled immunoglobulin as was labelled, but
191 10000
y////~
Y////,
Y////~
80OO
////// ////// //////
////// /i//// //////
6OOO o
/////I //////
o
4000
I///11 I/ // // /f /l /l I////I IIIIIi //11/l ~111.. IIIIII ////// ////// //////
2000-
~ *Rb.
U,/,/,/,. /x///x/ // x/ /
V/Ill,
/ / / / / / ~
I///I/ //////
ffZ777-~...... t111/i /////i
"Rb.+Rb.
*MA.
vzzg/( "MA+MA
*R
*R+R
"MCG "MCG+MC
*Fc
FC
1
*F(ab)2
22,500CPMOFANTIBODYPROBEADDED
Fig. 1. This graph shows the binding of various 12sI-immunoglobulin probes to 1.8 mg of Ps. maltophilia, and their competitive displacement with unlabelled immunoglobulins of various species. Note that 12SI-radiolabelled F(ab') 2 of IgG did not bind to the bacteria. (a) * Rb = 12SI-labelled polyclonal, rabbit anti-CG. (b) * Rb + Rb = x25I-labelled rabbit anti-CG in presence of 100 txg of unlabelled rabbit anti-CG. (c) * MA = 12SI-labelled monoclonal, mouse anti-c~-CG. (d) * MA + MA = 125I-labelled mouse anti-a-CG in presence of 10/xg of unlabelled mouse anti-a-CG. (e) * R = tzsI-labelled polyclonal, rabbit anti-goat IgG. (f) * R + R = tzSI-rabbit anti-goat IgG in presence of 10/xg of unlabelled rabbit anti-goat IgG. (g) * MCG = t2SI-labelled mouse anti-CG. (h) * MCG + MCG = lzSI-labelled monoclonal, mouse anti-CG in presence of 10/xg of unlabelled mouse anti-CG. (i) * Fc = 12sI-labelled Fc of IgG. (j) Fc = 12SI-labelled Fc of IgG in presence of 10/xg of unlabelled Fc of IgG. (k) * F ( a b ' ) 2 = 12SI-labelled F(ab') 2 of IgG.
also by unlabelled immunoglobulin from a different species. Thus, in Fig. 1 are shown the binding and competitive displacement of labelled rabbit, and mouse antibodies and of labelled Fc of IgG. This binding and displacement did not relate to the antigenic determinants recognized by the immunoglobulins, but only to the fact that they were immunoglobulins. Thus, both rabbit anti-CG and rabbit anti-goat immunoglobulin bound and were competitively displaced. Similarly, both mouse monoclonal anti-CG and anti-a-glycoprotein subunit bound and were competitively displaced. The Fc portion of human IgG also bound and was competitively displaced. Unlabelled F(ab') 2 fragments did not show binding.
Immunoglobulin dose-response curves with whole bacteria The next set of studies was performed by incubating 1.8 mg of whole bacteria with the five ~2SI-radiolabelled antibodies, in the presence of increasing amounts of the respective unlabelled immunoglobulin. The percent labelled immunoglobulin increased from an average of 8 15% when the amount of bacteria was doubled from 0.9 to 1.8 mg per tube. For all five antibod-
ies (two mouse, two rabbit and the Fc portion of human IgG), there was a dose-response related competitive displacement of the 12SI-labelled immunoglobulin or Fc fragment (data not shown). The amount of immunoglobulin producing maximal displacement of label varied as follows: human F c y = 1.8 /zg, rabbit anti-CG = 1 ~g, rabbit anti-goat IgG = 10 ~g, mouse anti-CG = 10/xg.
Binding studies using partially purified Fc-binding protein The sonicate supernate prepared as described in the materials and methods section was partially purified by serial Sephacryl S-300 followed by Sephadex G-100 column chromatography. The immunoglobulin binding property of the eluted fractions of both the Sephacryl S-300 and Sephadex G-100 showed that it was dispersed over a range of 85 kDa to 60 kDa (data not shown). We observed that there was an aggregation of the pure protein, after the labelled immunoglobulin was added to it, thus, protein is forming a much larger complex with the 125I-Fc fragment, making it large enough to be centrifuged at 10,000 x g, the step used to separate bound and free 125I-Fcy.
192 MYELOMA Fc/F(ab)2
120
120"
•
A---O-.dk ~--
•
[]
Prot.A.(Fc)
•
¢--.__~ el ', a ~ . . _ . , ~ D
100
IgA
0
F(=b)2
o
Prot.G.(Fc)
A
P.Malt.(Fc)
[]
ProtJ,.(F(ab))
•
Prot.G.(F(ab))
•
P,Malto.(F(ab))
100
0 o
~,A
°o
o
•
Prot. A,
o
Prot. G.
A
P. Malto.
•
Prot. A.
o
Prot. G.
&
P. Malto.
\\ 80
80
o<
'ii,
O ¢D
40
o-e o co
:Z
so
60-
co
',x,
.
40,
',\ \11"~ "\"A
ix
20
oN
o~
A 0
B
" ~ O~o
........
.01
FC
"R
\ ,
........
,
.1
.......
1
~
10
.......
,
100
.0001
.001
.01
.1
1
10
DOSE OF IMMUNOGLOBULIN (~lg)
DOSE OF IMMUNOGLOBULINFRAGMENT(pg)
IgG2 IgG1 120"
120
100
100
\ \
,,
80
•
Prot.A.
o
Prot.G.
A
P. Malto.
\\,.
80"
i~
o-e O
m
60
40 l'i
60o CD
40-
20'
20
C •0001
"~~---8.001
.01
.1
1
10
DOSE OF IMMUNOGLOBULIN FRAGMENT(pg)
.0001 .001
.01
.1
1
10
100
DOSE O F I M M U N O G L O B U L I N FRAGMENT(~Jg)
Fig. 2 (continued on next page)
193
IgG4
IgG3 120"
120 O A,
• •
•
100
100
Prot.A.
~o.~
•
\
o Prot.G. 01
80
o~
A
l
I 1
& P. Malto.
',
Prot.A.
o Prot.G. P. Malto.
so. I/
'1
o
0
60"
m
40"
!I
I
\\, 2o.
"~
"..o~_,
E
.001
.01
.1
1
10
DOSE OF IMMUNOGLOBULIN FRAGMENT(pg)
o
...................................
0"s 1 0"4 1 0"3 10"2 10"1 1 0 ° 101
DOSE OF IMMUNOGLOBULIN FRAGMENTgJg )
Fig. 2. These graphs show the binding and competitive displacement of 125I-labelled Fc of IgG by increasing doses of unlabelled Fc, myeloma IgA and various immunoglobulin subclasses on incubation with 1 /~g of the partially purified Pseudomonas protein. The cpm bound are all normalized to 100%, defined as cpm bound in tubes containing no added unlabelled immunoglobulin or immunoglobulin fragments. All the graphs show a dose-response relationship except for F(ab') 2 which does not compete for binding of 125I-Fc. Staphylococcal protein A and streptococcal protein G are run alongside for each immunoglobulin understudy as a comparison. A: dose-response curve showing binding of 125I-Fc7 with its competitive displacement by unlabelled Fc3, and F(ab') 2 with the Pseudomonas partially pure protein, protein A and protein G. B: dose-response curve showing binding of 125I-Fc-/ with its competitive displacement by increasing doses of unlabelled murine monoclonal IgA. C: dose-response curve showing binding of 12SI-Fcy with its competitive displacement by unlabelled human IgG1. D: dose-response curve showing binding of 12Sl-Fc7 with its competitive displacement by increasing doses of unlabelled human IgG2. E: dose-response curve showing binding of 125I-Fc7 with its competitive displacement by increasing doses of unlabelled human IgG3. F: dose-response curve showing binding of ~251-Fc7 with its competitive displacement by increasing doses of unlabelled human IgG4.
Dose-response relationships were observed when the partially pure protein was incubated with 125I-Fc of human IgG, and increasing doses of unlabelled Fc of human IgG, (F(ab') 2 of human IgG, murine monoclonal IgA, and human IgG1, ]gG2, IgG3, IgG4. All the immunoglobulin fractions displaced radiolabelled Fc except F(ab') 2 of IgG. These data are shown in Figs. 2A-2F. Similar dose-response relationships were also observed when protein A of Staphylococcus or protein G of Streptococcus was used as binding proteins. For all the above studies the concentra-
tions of protein A, protein G, and the Pseudomonas protein were normalized by first performing titres to give 25-27% binding of tzsI-Fc. With two exceptions, all immunoglobulins competed for tzSI-Fc binding in identical dose response relations for all three immunoglobulin binding proteins. For protein G, murine monoclonal IgA failed to compete for Fc binding, while for Pseudomonas protein, IgA inhibited Fc binding. For staphylococcal protein A, IgG3 failed to compete for Fc binding, while for Pseudomonas protein, IgG3 inhibited Fc binding. Unlabelled human F(ab') 2 did not compete for label binding.
194 100.
TABLE I EQUILIBRIUM CONSTANTS OF VARIOUS PROTEINS U S I N G H U M A N IgG
8,.I
Equilibrium constant ( h u m a n IgG) Protein A a Protein G a Pseudomonal b Ig-binding protein
6.7× 10 -1° 4.4× 10 -1° 1.54 × 1 0 - 1 ° / 2 . 3 6 × 10-8 c
a Akerstrom and Bjorck, 1986. Determined by Scatchard analysis using h u m a n IgG coupled to Sepharose beads and radioiodinated p s e u d o m o n a s Ig-binding protein. c Two slopes were obtained in the Scatchard plot.
The control proteins, BSA and transferrin, showed no binding of labelled immunoglobulin (data not shown). The kinetics of I g G - P s e u d o m o n a s Ig-binding protein was used using human I g G coupled to Sepharose. Our initial time studies showed that the binding is slow, reaching 50% after nearly 10 h. From here the binding was even slower, saturating at 18-20 h. This interaction was studied at 4 ° C and 37 ° C, using h u m a n IgG. The binding at 4 ° C greatly decreased, taking nearly 18 h to achieve 50%. Table I compares the equilibrium constants of the Pseudomonas Ig-binding protein with that of protein A and protein G. As the equilibrium constant ( K d) of the reaction is equal to the absolute value of the slope of the curve, we identified two distinct types of Ig-binding proteins which are present in Pseudomonas. One has a very high affinity ( K d = 1.545 × 10-10) and one with a low affinity ( K d = 2.36 × 10-8). As seen in Table I, one type of the p s e u d o m o n a s Ig-binding protein has an equilibrium constant in the same magnitude as that of protein A and protein G. There also exists a binding protein on Pseudomonas of very low affinity. The p r e p o n d e r a n c e of which may explain the length of time (10 h) required to achieve 50% saturation as c o m p a r e d to protein A, protein G. Fig. 3 shows the Scatchard diagram of the Ig-binding protein, using human I g G coupled to Sepharose and radiolabelling the Pseudomonas Ig-binding protein. H e r e two slopes were obtained: one having a high affinity ( K d = 1.545 x 10 - l ° ) and one with a low affinity (K~ = 2.36 × 10-8).
.545~10-10 6o
I1 m 4o,
1I 1
""" 19
39
59
79
99
119139159179199
BOUND (P. MOL.) Fig. 3. The figure shows the Scatchard plot for the reaction between the pseudomonas Ig-binding protein and human IgG coupled to Sepharose beads. Here the equilibrium constant (K d) of the reaction is equal to the absolute value of the slope of the curve. Two different types of Ig-binding protein were seen: one with high affinity (K d = 1.545 × 10-10) and one with low affinity (K d = 2.36× 10-8).
PAGE / Western blot studies Fig. 4A shows a 12% polyacrylamide gel stained with Coomassie blue. Lane 1 shows the profile of the sonicated proteins. Lane 2 shows the profile of the purified preparation. Note that there is a major band at 30 kDa and there are also minor bands at 69 and 5 kDa. Fig. 4B shows Western blot analysis of the purified protein hybridized with various radiolabelled probes. As seen in Fig. 4B, F c y murine monoclonal IgA, various h u m a n i m m u n o g l o b u l i n subclasses showed reaction towards the immobilized protein, whereas human F(ab') 2 did not show any reaction. The major band identified in Fig. 4A, lane 2, was identified to be the immunoactive part at 30 kDa.
Discussion Proteins, with affinity for different regions of mammalian IgG, have been isolated and well characterized from different strains of bacteria. The ability of Protein A, derived from Staphylo-
195
MOL WT Kd
I
2
A)
93B) MOL WT
69-
Kd
I
2
3
4
5
6
7
4593--
3069-45--
2130--
14.3-
21~
14.3--
_
5--
Fig. 4. A: 25 /.~g of the sonicated and purified material were electrophoresed on a 12% polyacrylamide gel, and stained with Coomassie blue R-250. Lane 1 shows the profile of the sonicated protein with a wide range of molecular weights, whereas, after purification there is a large prominent band at 30 kDa region with minor bands at 69 kDa and 5 kDa regions. B: the figure shows the Western blot analysis of the purified Ig-binding material, hybridized for 24 h, with the following radiolabelled human immunoglobulin fragments and subclasses. Lane 1, x25I-Fcy; lane 2, 125I-murine monoclonal IgA; lane 3, 125I-IgG1; lane 4, 125I-IGG2; lane 5, 125I-IGG3; lane 6, 125I-IGG4; lane 7, 1251-F(ab')2.
coccal aureus, to bind IgG via its Fc region has made it an important immunological tool. In addition, streptococci groups A and C possess a surface protein which binds the Fc regions of IgG (Kronvall, 1973), and groups C and G have surface proteins which bind the F(ab') 2 regions of IgG (Kronvall et al., 1983). Very recently Lindahl and Akerstrom (1989) have described the existence of an IgA binding protein in group A Streptococcus. Strains of Streptococcus and Peptococcus magnus have also been identified which possess surface proteins, which preferentially bind light chains of immunoglobulins (Bjorck et al., 1986; Bjorck, 1988). We have also demonstrated that Ps. maltophilia also has an immunoaccessible surface protein which binds IgG via their Fc regions, murine monoclonal IgA, but not F(ab') 2 of IgG. As described in the materials and methods section, we could not detect any contamination of the human IgG subclasses with small amounts of human IgA by the Western blot method. We also eliminated the possibility of serum IgA anti-pseudomonal antibodies giving
false binding in our studies by using IgA secreted by an ATCC myeloma cell line which secrets only monoclonal IgA antibodies directed against trinitrophenol. Thus, the binding of IgA by the Pseudomonas protein appears to be a valid observation. We observed a discrepancy between the molecular weight of this protein as estimated by Western blot analysis which gave a weight of 30 kDa, and by the elution profile of the immunoglobulin binding property of Sephacryl S-300 and Sephadex G-100 chromatography, which gave a weight range of 60-80 kDa. Here the elution profile was polydispersed over a wide range of tubes rather than a sharp peak (data not shown). This could be possible if the protein molecule is aggregating into large polymeric complexes or that it is associated with a heavier protein. The gel electrophoresis is performed under denaturing conditions, thus, the aggregated protein could be denatured into its individual components, one of which is immunoactive. We tested for presence of Fc binding activity in the supernate of the partially pure protein after centrifugation at
196
10,000 x g and could detect none. This finding would also suggest that the native protein may exist in an aggregated form or associated with a larger MW protein. We observed large insoluble complexes with polyclonal IgG. This would indicate that the Ig-binding protein has several binding sites akin to protein G which has at least 2/molecule and protein A, 4/molecule. Scatchard analysis showed that there are two types of the protein, one with high affinity (K d = 1.545 × 10 -1°) and one with low affinity (K d = 2.36× 10 8), when using human IgG. The low affinity may exist in much larger quantity, thus explaining the 20 h period of incubation required to attain saturation of binding to IgG-Sepharose. This Pseudomonas protein differs in the spectrum of immunoglobulins bound from that of staphylococcal protein A in that it binds human IgG3, while protein A does not bind IgG3. The pseudomonal immunoglobulin binding protein also differs from protein G in that the former binds IgA while protein G does not bind IgA. These properties suggest the three binding proteins are products of different genes. We discovered the Fc-binding protein in the course of studies designed to evaluate reports that a CG-like protein was produced by Pseudomonas. Acevedo et al. (1978) have reported this observation based on anti-CG antibody binding to Pseudomonas. Richert and Ryan (1977) have reported that CG-like receptors exist on the surface of Ps. mahophilia, a finding that does suggest CG-like materials may effect Pseudomonas actions, growth or some other function. In lieu of the presence of immunoglobulin binding proteins on the surface of Ps. maltophilia, it is difficult to interpret antibody binding identification of CG in Pseudomonas. We found considerable artifactual effects when CG immunoassays were used to study possible CG production by Ps. maltophilia. The current studies indicate that the presence of any surface molecule on bacteria demonstrated by binding of free antibodies does not prove the presence of the molecule unless the studies also demonstrate that the antibody binding is specific and is not displaced by Fc fragments. Further, the current studies demonstrate an additional desirable feature of the use of gold
immunoprobes to explore bacterial surfaces. Gold immunoprobes, as conventionally constructed (Robinson et al., 1987), have the Fc portion oriented toward the center of the probe and display only or primarily the F(ab') 2 portions of bound antibodies. Thus, false positive binding of antibodies by their Fc portions (as demonstrated with free antibodies in the current paper) is sterically hindered. The wider spectrum of immunoglobulin classes bound by the Ps. mahophilia relative to staphylococcal protein A may make this protein a useful reagent in immunoabsorbent column chromatography and gold or silver immunoprobe construction.
Note added in proof
Pseudomonas maltophilia was transferred to the genus Xanthomonas as Xanthomonas maltophilia (Hugh 1981) (Swings, J., De Vos, P., Van der Nooter, M. and De Ley, J. (1983) Int. J. Syst. Bacteriol. 33, 409).
Acknowledgements We would like to thank Gary Gorby, Jeanine Griffin, and John Klein for their technical assistance and cooperation. We thank the late Virgie Roubidoux and Char Kidd for editorial assistance and preparation of this manuscript. This paper is based upon work supported by National Institutes of Health Grant no. RO1 HD18986.
References Acevedo, H.F., Slifkin, M., Pouchet, G.R. and Pardo, M. (1978) Immunohistochemical localization of choriogonadotropin-like protein in bacteria isolated from cancer patients. Cancer 41, 1217. Bjorck, L. (1988) Protein L, A novel bacterial cell-wall protein with affinity for Ig L chains. J. Immunol. 140, 1194. Bjorck, L. and Akerstrom, B. (1986) A physiochemical study of Protein G, a molecule with unique immunoglobulin G-binding properties. J. Biol. Chem. 261, 10240. Bjorck, L., Logdberg, L., Berggard, B., Schalen, C. and Persson, M.H. (1986) Related bindings of aggregated /32-mi-
197 croglobulin, IgG Fab, kappa and lambda light chains to group A Streptococci. Acta Pathol. Microbiol. Immunol. Scand. 94, 25. Enderson, C. (1979) The binding to Protein A of immunoglobulin G and of Fab and Fc fragments. Acta. Pathol. Microbiol. Scand. 87, 185. Fuller, S.A., Takahashi, M. and Hurrell, J.G.R. (1987) Purification of monoclonal antibodies. In: F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith and H. Struhl (Eds.). Current Protocols in Molecular Biology, Vol. II. Wiley, New York, pp. 11.11.1-5. Hjelm, H., Meloun, B. and Sjoquist, J. (1972) Protein A isolated from Staphylococcal aureus after digestion with lysostaphin. Eur. J. Biochem. 29, 572. Hunter, W.M. and Greenwood, F.C. (1962) Preparation of iodine 131 labelled human growth hormone of high specific activity. Nature 194, 495. Inganas, M. and Johansson, S.G.O. (1981) Influence of the alternative protein A interaction on the precipitation between human monoclonal immunoglobulins and protein A from Staphylococcus aureau. Int. Arch. Allergy Appl. Immunol. 65, 91. Kronvall, G. (1973) A surface component in group A, C, and G Streptococci with non-immune reactivity for immunoglobulin G. J. Immunol. 111, 1401. Kronvall, G. and Gunnar, L. (1988) Non-immune binding of Ig to clostridium perfringeus, preferential binding of IgM and aggregated IgG. J. Immunol. 140, 1223. Kronvall, G., Paul, G.Q. and Williams, Jr., R.C. (1970) Quantitation of Staphylococcal Protein A: determination of equilibrium constant and number of Protein A residues on bacteria. J. Immunol. 104, 273.
Kronvall, G., Myhre, E.B. and Erntell, M. (1983) Alternative non-immune F(abI)2 mediated immunoglobulin binding to group C and G Streptococci. Scand. J. Immunol. 17, 201. Langone, J.J. (1982) Protein A of Staphylococcus aureus and related immunoglobulin receptors produced by streptococci and pneumococci. Adv. Immunol. 32, 157. Lindahl, G. and Akerstrom, B. (1989) Receptor for IgA in Group A Streptococci: cloning of the gene and characterization of the protein expressed in Escherichia coli. Mol. Microbiol. 3, 239. Lowry, O.H., Rosebrough, N.J. Farr, A.L. and Randall, R.J. (1951) Protein measurement with the folin phenol reagent. J. Biol. Chem. 193, 265. Odell, W.D. and Heber, D. (1980) Radioiodination of peptides and proteins for radioimmunoassays or radioreceptor assays. In: N.R. Rose and H. Friedman (Eds.), Manual of Clinical Immunology, 2nd edn. American Society for Microbiology, Washington, DC, p. 329. Richert, N.D. and Ryan, R.J. (1977) Specific gonadotropin binding to Pseudomonas maltophilia. Proc. Natl. Acad. Sci. U.S.A. 74, 878. Robinson, Jr., E.N., McGee, Z.A. and Clemens, C.M. (1987) Probing the surface of bacteria: use of gold sphere immunological probes. Microbiol. Pathogen. 2, 159. Scatchard, G. (1949) The attractions of proteins for small molecules and ions. Ann. N.Y. Acad. Sci. U.S.A. 51, 660. Towbin, H., Staehlin, T. and Gordon, J. (1979) Electrophorectic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. U.S.A. 76, 4350.