Binding Kinetics of Antiricin Single Domain Antibodies and Improved

Aug 5, 2010 - attractive alternatives to conventional monoclonal anti- bodies. ... domain antibody binding kinetics using surface plasmon resonance an...
3 downloads 0 Views 3MB Size
Anal. Chem. 2010, 82, 7202–7207

Binding Kinetics of Antiricin Single Domain Antibodies and Improved Detection Using a B Chain Specific Binder George P. Anderson,† Rachael D. Bernstein,‡ Marla D. Swain,† Dan Zabetakis,† and Ellen R. Goldman*,† Center for Bio/Molecular Science and Engineering, Naval Research Laboratory, 4555 Overlook Avenue SW, Washington, D.C. 20375, and Nova Research Inc., 1900 Elkin Street, Suite 230, Alexandria, Virginia 22308 Single domain antibodies are the recombinantly expressed binding fragments derived from heavy chain antibodies found in camels and llamas. These unique binding elements offer many desirable properties such as their small size (∼15 kDa) and thermal stability, which makes them attractive alternatives to conventional monoclonal antibodies. We created a phage display library from llamas immunized with ricin toxoid and selected a number of single domain antibodies. Phage selected on ricin were found to bind to either ricin A chain or the intact molecule; no ricin B chain binders were identified. By panning on B chain, we identified binders and have characterized their binding to the ricin B chain. While they have a poorer affinity than the previously described A chain binders, it was found that they performed dramatically better as capture reagents for the detection of ricin, providing a limit of detection in enzyme linked immunosorbent assay (ELISA) below 100 pg/mL and excellent specificity for ricin versus the highly related RCA 120 (1 to 10 000). We also reevaluated the previously isolated antiricin single domain antibody binding kinetics using surface plasmon resonance and found their Kds matched closely to those previously obtained under equilibrium binding conditions measured using the Luminex flow cytometer.

beans for their oil, the ease of ricin isolation from the waste, and its high stability makes ricin a considerable biothreat,1 although fortunately, like most biothreats, not one trivial to implement.5 The development of sensitive sandwich immunoassays for nonrepetitive targets, such as protein toxins like ricin, requires high affinity antibodies that bind to different locations on the target. Previously, we reported on the development of a phage display library of single domain antibodies (sdAb) derived from llamas that had been immunized with ricin toxoid. Llamas and the other members of the Family Cameledia have unique immune systems, which include heavy chain only antibodies.6 These antibodies lack light chains; thus, antigen binding is mediated by only the variable heavy domain.7 SdAb are the recombinantly produced variable domains from these heavy chain antibodies.8-10 Taking advantage of modern molecular biology techniques, the immune repertoire of the heavy chain antibodies can be cloned into a phage display vector, which allows the wide assortment of heavy chain variable domains to be expressed on the end of a minor coat protein of a filamentous phage.11-13 Using panning techniques allows one to isolate and amplify the binders and eventually derive monoclonal sdAb that have affinity toward the target of interest. Our earlier work isolated and evaluated a number of sdAb which bound ricin, most with high affinity and specificity.14 The

Ricin is a highly toxic 60 kDa protein which is found in the castor bean. Ricin is the most common member in the family of type 2 ribosome inhibiting proteins (RIPs). All type 2 RIPs function as AB toxins, where the B subunit facilitates cell binding and entry, while the A subunit possesses the enzymatic activity, hydrolyzing the N-glycosidic bond of a particular adenine within the 28S rRNA, which inhibits protein synthesis and ultimately causes cell death.1-4 Since a single molecule of ricin can induce cell death, the toxin is very potent. The large worldwide production of castor

(5) Schep, L. J.; Temple, W. A.; Butt, G. A.; Beasley, M. D. Environ. Int. 2009, 35, 1267–1271. (6) Hamerscasterman, C.; Atarhouch, T.; Muyldermans, S.; Robinson, G.; Hamers, C.; Songa, E. B.; Bendahman, N.; Hamers, R. Nature 1993, 363, 446–448. (7) Arbabi Ghahroudi, M.; Desmyter, A.; Wyns, L.; Hamers, R.; Muyldermans, S. FEBS Lett. 1997, 414, 521–526. (8) Muyldermans, S. Rev. Mol. Biotechnol. 2001, 74, 277–302. (9) Muyldermans, S.; Baral, T. N.; Retarnozzo, V. C.; De Baetselier, P.; De Genst, E.; Kinne, J.; Leonhardt, H.; Magez, S.; Nguyen, V. K.; Revets, H.; Rothbauer, U.; Stijemans, B.; Tillib, S.; Wernery, U.; Wyns, L.; HassanzadehGhassabeh, G.; Saerens, D. Vet. Immunol. Immunopathol. 2009, 128, 178– 183. (10) Wesolowski, J.; Alzogaray, V.; Reyelt, J.; Unger, M.; Juarez, K.; Urrutia, M.; Cauerhff, A.; Danquah, W.; Rissiek, B.; Scheuplein, F.; Schwarz, N.; Adriouch, S.; Boyer, O.; Seman, M.; Licea, A.; Serreze, D. V.; Goldbaum, F. A.; Haag, F.; Koch-Nolte, F. Med. Microbiol. Immunol. 2009, 198, 157– 174. (11) Smith, G. P.; Scott, J. K. Methods Enzymol. 1993, 217, 228–257. (12) Scott, J. K.; Smith, G. P. Science 1990, 249, 386–390. (13) McCafferty, J.; Griffiths, A. D.; Winter, G.; Chiswell, D. J. Nature 1990, 348, 552–554. (14) Anderson, G. P.; Liu, J. L.; Hale, M. L.; Bernstein, R. D.; Moore, M.; Swain, M. D.; Goldman, E. R. Anal. Chem. 2008, 80, 9604–9611.

* Corresponding author. Phone: 1-202-404-6052. Fax: 1-202-767-9594. E-mail: [email protected]. † Naval Research Laboratory. ‡ Nova Research Inc. (1) Audi, J.; Belson, M.; Patel, M.; Schier, J.; Osterloh, J. JAMA, J. Am. Med. Assoc. 2005, 294, 2342–2351. (2) Kozlov, Y. V.; Sudarkina, O. Y.; Kurmanova, A. G. Mol. Biol. 2006, 40, 711–723. (3) Olsnes, S. Toxicon 2004, 44, 361–370. (4) Barbieri, L.; Battelli, M. G.; Stirpe, F. Biochim. Biophys. Acta 1993, 1154, 237–282.

7202

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

10.1021/ac100961x  2010 American Chemical Society Published on Web 08/05/2010

approximate Kd was determined by monitoring the equilibrium binding to toxin coated microspheres. Then, using competitive binding assays, three different epitopes were defined, all putatively on the ricin A chain. Using a sdAb capture/tracer pair that recognizes two different epitopes, an assay sensitive to 1.6 ng/mL ricin was demonstrated. Using the sdAb displayed on the phage as the tracer molecule, it was possible to push that sensitivity even further to subnanogram per milliliter levels.12,13,15,16 Most of the initial work focused on only three of the sdAb out of the nine sequence families that were initially isolated. Those three sdAb were chosen both for their good affinity and the fact that each recognized a unique epitope, facilitating their use when paired for sandwich assays. However, we still desired to obtain binders toward the ricin B chain. We had originally isolated two sdAb by panning on B chain, but our initial work on those binders was not promising nor were they found to bind significantly to ricin B chain immobilized on microspheres. Here, we have isolated additional binders by selecting on ricin B chain and have evaluated both new and previously isolated sdAb by surface plasmon resonance (SPR) to determine affinity constants and to better define the binding epitopes of the sdAb. By taking advantage of the ProteOn SPR’s six by six array, we were able to simultaneously examine the affinity constants toward ricin and the related molecule RCA 120, as well as ricin A chain and ricin B chain. Cross reactivity of antibodies to RCA 120 are a good test of specificity, since it shares 80% homology to ricin but is a relatively nontoxic tetramer, as opposed to ricin which is a highly toxic dimer.17 Herein, we describe the isolation and characterization of ricin B chain binding sdAb. In addition, we developed a more complete description of kinetic binding parameters and binding epitopes for all the antiricin sdAb. We also describe the development of a highly sensitive and specific immunoassay by pairing an antiricin B chain sdAb with an antiricin A chain sdAb. MATERIALS AND METHODS Reagents. Ricinus communis agglutinin II (ricin), ricin A chain, ricin B chain, Ricinus communis agglutinin I (RCA120), and goat antiricin were purchased from Vector (Burlingame, CA). Ricin toxoid was from Toxin Technologies (Sarasota, FL). The Mabs Ric03AG1 and Ric07AG1 were the kind gift of Dr. Jill Czarnecki (Naval Medical Research Center, Silver Spring, MD). Llama immunizations of two animals were through triple J farms (Bellingham, WA). PhycoLink Streptavidin-R-Phycoerythrin PJ31S (SA-PE) and Streptavidin-Horseradish Peroxidase (SA-HRP) were provided by Prozyme (San Leandro, CA). Phosphate buffered saline (PBS), Tween 20, and bovine serum albumin (BSA) were obtained from Sigma-Aldrich (St. Louis, MO). SdAb Panning, Production, and Biotinylation. Selection was carried out as previously described using ricin B chain as the target.14 Individual binding clones were identified through monoclonal phage enzyme linked immunosorbent assay (ELISA) which were performed after two rounds of panning. Selected positive clones were sequenced to identify unique sdAb genes. (15) Goldman, E. R.; Liu, J. L.; Bernstein, R. D.; Swain, M. D.; Mitchell, S. Q.; Anderson, G. P. Sensors 2009, 9, 542–555. (16) Goldman, E. R.; Anderson, G. P.; Bernstein, R. D.; Swain, M. D. J. Immunol. Methods 2010, 352, 182–185. (17) Roberts, L. M.; Lamb, F. I.; Pappin, D. J. C.; Lord, J. M. J. Biol. Chem. 1985, 260, 5682–5686.

Unique sdAb clones were subcloned from the phage display sdAb-fusion vector to a soluble sdAb expression vector and constructs were transformed into E. coli rosetta (Novagen, Madison, WI) for protein production. As described previously, the sdAb proteins were isolated from the periplasmic compartment of 500 mL scale shake flask cultures by osmotic shocking, immobilized metal affinity chromatography (IMAC), and gel filtration on a Superdex G75 column (GE-Healthcare).18 All the sdAb preparations eluted from the G75 column at volumes consistent with their being monomeric. Proteins were quantified using a microbicinchoninic acid (BCA) assay (Pierce, Rockford, IL) and stored at 4 °C prior to analysis. For use as tracer reagents in both Luminex and ELISA assay, both sdAb and convention antibodies were labeled using biotin (Bt)-LC-LC-NHS from Pierce. The Bt-LC-LC-NHS was dissolved in DMSO and then added to the antibody in PBS at a molar ratio of 10 to 1. After 30 min, the biotinylated antibody was separated from the free biotin by gel filtration (Biogel P10, Bio-Rad, Hercules, CA). Preparation of Luminex Reagents and Assay Protocols. Luminex (Austin, TX) xMAP carboxylated microspheres were cross-linked to a variety of proteins using the two-step carbodiimide coupling protocol provided by the manufacturer. Direct bindingassayswereperformedessentiallyasdescribedpreviously.14,18 Circular Dichroism (CD) Measurements. The melting point of the B4 sdAb was measured by circular dichroism using a Jasco J-815 CD spectropolarimeter equipped with a PTC-423S single position peltier temperature control system. Samples (∼30 µg/ mL) were prepared by extensive dialysis versus 5 mM sodium borate, pH 7.5. All measurements were made in a 10 mm path length quartz cuvette with a stir bar. The data were acquired from 245 to 195 nm at a scanning speed of 20 nm/min. The data pitch was 1 nm, DIT 2 s, bandwidth of 1 nm, and temperature ramp rate of 2 °C/min with a temperature interval of 5 °C over the range of 25-85 °C. ELISA. ELISAs were performed essentially as described previously.16 Capture sdAb was immobilized on wells of 96-well maxisorb plates (Nunc) by incubating 100 µL of the sdAb diluted into PBS at 1 ug/mL overnight at 4 °C. Wells were washed with PBS with 0.05% Tween 20 (PBST) and blocked with 2% powdered milk in PBS. Dilutions of ricin or RCA120 were added to the wells in triplicate; wells containing just PBS were included as a no-antigen control. After washing with PBST, biotinylated sdAb reporter was added at 1 µg/mL. Wells were washed and then incubated for an hour with streptavidin-HRP. After a final washing with PBST, signal was developed by the addition of sigmafast OPD (Sigma). The ELISA testing for ricin binding in 4% milk was performed the same, except the ricin was diluted into a solution of 4% powdered milk that was made up in PBS and wells containing only the milk solution were used as controls. SPR Kinetics Analysis. The SPR kinetic measurements were performed using the ProteON XPR36 (Bio-Rad). For testing the kinetics of the antiricin sdAb, a GLC chip was coated with the ricin (5 and 3 µg/mL) along with RCA 120 (5 and 3 µg/mL) and ricin A and B chains each at 5 µg/mL. Other tests utilized sdAb specific for ricin immobilized to a GLC chip to measure the (18) Goldman, E. R.; Anderson, G. P.; Liu, J. L.; Delehanty, J. B.; Sherwood, L. J.; Osborn, L. E.; Cummins, L. B.; Hayhurst, A. Anal. Chem. 2006, 78, 8245–8255.

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

7203

Table 1. Determination of KD Parameters of the Antiricin sdAba

a Earlier KD values determined by equilibrium binding measurement using a Luminex flow cytometer are compared to KD values obtained on the ProteOn SPR. KD values for binding to RCA120 and ricin A chain were also obtained. The relative increase in maximum resonance units (RU) for RCA120 and ricin A chain compared to ricin is also given. LB: low binding observed, no binding parameters determinable. NB: no binding observed.

binding to antigen in solution. For immobilization, proteins were diluted in 10 mM acetate buffer pH 5.0 and attached to the chip following the standard EDC coupling chemistry provided by the manufacturer. All experiments were performed at 25 °C. The binding of the sdAb were tested by flowing six concentrations varying from 30 to 0 nM at 50 µL/min for 180 s over the antigen coated chip and then monitoring dissociation for 900 s. The chip was regenerated using 50 mM glycine-HCl (pH 2.0) for 36 s, prior to any additional testing. The data were analyzed with the ProteON Manager TM 2.1 software, corrected by subtraction of the zero antibody concentration column as well as interspot corrected; the binding constants were determined using the software’s Langmuir model. Assays for binding of ricin, RCA120, ricin A chain, and ricin B chain by the immobilized sdAb were done in an analogous manner. RESULTS AND DISCUSSION Ricin A chain is highly immunogenic and alone can produce a protective response.19 In our earlier work, we described the development of sdAb derived from an immune library that recognized intact ricin and the ricin A chain. Using equilibrium binding assays on the Luminex flow cytometer, we estimated the Kd of 16 isolated sdAb as well as two monoclonal antibodies (Table 1). Performing sandwich immunoassays with a more limited set of binders, we were able to ascertain that the sdAb recognized at least three different epitopes that were either on the A chain or favored the intact molecule, but we were not clearly able to ascertain whether any of the initial binders selected were B chain binders.14 This was in spite of the fact that two of the binders, B5H and B10E, had both been selected on the ricin B (19) Vitetta, E. S.; Smallshaw, J. E.; Coleman, E.; Jafri, H.; Foster, C.; Munford, R.; Schindler, J. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 2268–2273.

7204

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

chain. Since B chain, while not reliably protective, is know to be immunogenic and others have obtained B chain monoclonals, we repeated the selection of binders on ricin B chain and obtained two binding sequences: B4, which was identical to the previously isolated clone B10E, and B5, which was very similar to B10E but had several point mutations. The amino acid sequences of these sdAb are shown in Figure S1, Supporting Information. Recently, our laboratory obtained a ProteOn XPR36, and we have reevaluated the binding kinetics and specificity of all the antiricin sdAb by SPR. The Kd determined by SPR matched well with the results previously obtained by the Luminex equilibrium measurements (Table 1). A few of the sdAb performed much better when measured by SPR, which we attribute primarily to improved quality of purified sdAb. Clone E9 is the best example of this, as it gave very poor results in the initial tests but performed well when measured by SPR; on the basis of it’s high homology to other sdAb with low Kds, it was not surprising for E9 to have done so well. We had suspected some of the initially determined Kds to have large errors, which was why only a range of affinities, as opposed to the value for each clone, was originally reported. We found that sdAb within several sequence families, including the ricin B chain binders and the D1/F11 family (shaded green and yellow in Table 1 respectively), showed identical Kds despite the fact that the ricin B chain binders have nine differences between the clones and the D1/F11 have five sequence changes. Related clones in other families (those shaded in blue and pink in Table 1) showed a range of Kds as determined by SPR. For example, clones C2 and H3 differ in only two amino acid positions, but their Kds were calculated at 6 × 10-10 and 6.5 × 10-11 M, respectively. Similarly, the related clones F8 and H1 have 13 sequence differences and Kds that differ by an order of magnitude. Although it is possible for point mutations to have dramatic effects on the affinity of an antibody fragment, our ability to discern the difference between 10-10 and 10-11 M in the SPR experiments is dependent on the measurement of slight differences in off rates. Differences measured between high affinity clones may not be as significant as those measured between lower affinity binders whose off rates are easily measurable. The ProteOn XPR36 enabled us to evaluate the binding of the sdAb to ricin while simultaneously evaluating their binding to RCA120 as well as ricin A chain (Table 1) and ricin B chain (Table 2 and Figure S2, Supporting Information). We had noticed during the fluid array assays that many of the sdAb bound RCA120 more weakly than ricin, but we had not examined the kinetics of that interaction. Most of the sdAb bound to RCA120 with an affinity 1 to 2 orders of magnitude weaker than to ricin (Table 1), primarily due to differences in their off rates. In addition to faster off rates, the total binding response measured was also dramatically reduced on RCA 120 versus ricin for many of the sdAb clones. However, two families of sdAb (D1,F11) and (F8,H1) bound to RCA120 to an equal extent, giving similar signals on both proteins but still binding RCA 120 with lower affinity. Interestingly, the two Mabs tested bound to both ricin and RCA120 with near identical affinities. Thus, one would not expect to be able to discriminate between ricin and RCA120 using the Mabs, whereas good discrimination using the sdAb should be possible. Binding

Table 2. Determination of ka (1/Ms), kd (1/s), and KD (M) for the Binding of Immobilized sdAb (B4, C8, F8, and F11) to Ricin, RCA120, Ricin A Chain, and Ricin B Chain in Solutiona ricin B4 C8 F8 F11

RCA 120

ricin A

ka

kd

KD

ka

kd

KD

3 × 106 2 × 106 2 × 106 8 × 105

2 × 10-3 1 × 10-5 2 × 10-5 3 × 10-5

9 × 10-10 8 × 10-12 1 × 10-11 3 × 10-11

3 × 105 2 × 105 3 × 106 2 × 106

2 × 10-3 4 × 10-4 3 × 10-6 2 × 10-4

6 × 10-9 2 × 10-9 1 × 10-12 9 × 10-11

ka

kd

ricin B KD

8 × 105

2 × 10-4

3 × 10-10

8 × 105

4 × 10-4

5 × 10-10

ka

kd

KD

2 × 106

4 × 10-3

2 × 10-9

a The affinity constants were determined from the binding of five concentrations of each antigen (30, 10, 3.3, 1.1, and 0.37 nM). Supplemental Figure 2 (Supporting Information) shows the SPR traces from which this data was determined.

Figure 1. Binding of sdAbs F11 and B4 (30, 10, 3.3, 1.1, and 0.37 nM) to ricin and RCA120 immobilized on a GLC chip measured by SPR. Results show how F11 binds both Ricin and RCA120 while B4 binds ricin, it binds RCA120 much less strongly.

to ricin B chain is not shown in Table 1, as it was not observed for any of the sdAb. The B chain was observed to immobilize, but the immobilized material was not recognized by any of the sdAb, including those selected on ricin B chain. This fact appears to be why we earlier were unable to identify any of the sdAb, including those selected on B chain, as ricin B chain binders. From examining the binding of sdAb to the immobilized antigens, we did find that sdAb B4, an sdAb selected on the ricin B chain, discriminated well between ricin and RCA120; thus, even though it was of lower affinity, we thought it might be of use in developing an assay specific for ricin. Figure 1 shows the binding differential of sdAb B4 to ricin and RCA120 and compares it to sdAb F11, which has relatively poor discriminatory abilities. The binding of sdAb to immobilized toxin was followed up by immobilizing the sdAb that bind to the three distinct epitopes on

ricin: C8, F8, and F11; along with B4, the putative B chain binder isolated from the most recent round of panning on B chain, onto a GLC-chip for SPR evaluation (Table 2, Figure S2, Supporting Information). Again, each sdAb was observed to bind ricin, with B4 having the lowest affinity. Additionally, in this format, each of these sdAb were found to bind to RCA120; however as before, C8 had an affinity 2 orders of magnitude lower and B4 had an affinity 1 order of magnitude lower than their binding to ricin. These results made this pair an attractive set to evaluate for specificity in a sandwich assay format. Immobilization of the sdAb also allowed better evaluation of binding to A and B chains, as the target is free in solution, not hindered by immobilization. In this format, one can easily measure the binding of B4 to ricin B chain. None of the other three sdAb exhibited any binding to the B chain. Clone C8 bound ricin A chain as expected from the result Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

7205

Table 3. Binding Inhibition of the sdAb as Determined by SPR Analysisa

a By definition, self inhibition was set to 100%. Inter sdAb inhibition was adjusted by the same factor. Four clear epitopes are evident. Both G5 and G7 overlap C8’s binding site but only partially inhibit each other. C12 and C3 also overlap other sdAb epitopes but are clearly not identical. G12 bound too poorly to ascertain with any certainty but did appear to overlap with C8’s epitope, but it was not quantified in this table.

Figure 2. ELISA based detection using sdAb B4 as the capture molecule and Bt-C8 (1 µg/mL) as the tracer molecule. Panel A compares the detection of ricin and RCA120. Panel B compares the detection of ricin diluted into PBS with ricin diluted into a 4% milk solution. Assay performed as described in the Materials and Methods. Error bars represent the standard deviation.

observed in Table 1. Both F11 and F8 bound minimally to immobilized A chain; thus, we were interested if the reverse configuration would be more enlightening. Clone F11 did prove to bind well to soluble ricin A chain, while clone F8 still bound the ricin A chain poorly, indicating that F8 favors the intact toxin, likely binding to an epitope composed of both A and B chains (Table 2, Figure S2, Supporting Information). These results show how immobilization of antigen or antibody can adversely impact the subsequent interaction measurement either due to the immobilization process causing denaturation of the bound molecule or causing steric hindrance due to favoring attachment in the wrong orientation. Our next interest was to reevaluate the number of binding epitopes represented by the different families of sdAb. Our earlier work had identified three clear epitopes for the families shown in Table 1 highlighted yellow, blue, and pink, which was the same number obtained by Maddaloni et al.20 This work identified a B chain binding family as the forth epitope, highlighted green. Epitopes had not been clearly determined for the sdAb that are not highlighted. To better investigate their binding epitopes, binding inhibition by each sdAb for the others was evaluated by SPR (Table 3). This analysis still yielded only four clearly separated epitopes; the other sdAb were found to bind to overlapping but nonidentical epitopes. Both G5 and G7 overlap C8’s binding site but only partially inhibit each other. C12 and (20) Maddaloni, M. C., C.; Wilkinson, R.; Stout, A. V.; Eng, E.; Pincus, S. H. J. Immunol. 2004, 172, 6221–6228.

7206

Analytical Chemistry, Vol. 82, No. 17, September 1, 2010

C3 also over lap other sdAb epitopes but are clearly not identical. G12 bound too poorly to ascertain with any certainty but did appear to overlap with C8’s epitope. Having identified a new ricin B chain binding sdAb that had little recognition of RCA120, we investigated how well it would function as a capture molecule in both ELISA and fluid array immunoassays. The ELISA results were most impressive, using B4 as a capture molecule and Bt-C8 as the tracer molecule gave a limit of detection of less than 100 pg/mL and a specificity of ricin over RCA120 of greater than 1 to 10 000 (Figure 2a). Although the amount of cross reactivity to RCA 120 at the highest concentrations does vary between assays, we always achieved discrimination of at least 1 to 1000. We have also performed ELISA assays for the detection of ricin spiked into milk (4% evaporated milk; Figure 2b). Not unexpectedly, the limits of detection are about 25- to 100-fold less sensitive; limits of detection are often worse in complex matrixes.21,22 On the basis of the relatively low affinity of B4 for ricin due to its high off rate, it functioned surprisingly well as a capture molecule in the ELISA format. One possible reason is that by binding ricin’s B chain it leaves C8’s epitope on the A chain easily accessible, and obviously, this format is not as impacted by B4’s off rate as one might presuppose. Perhaps the ELISA format favors rebinding to the surface, thereby (21) Garber, E. A. E.; Venkateswaran, K. V.; O’Brien, T. W. J. Agric. Food Chem. 2010, 58, 6600–6607. (22) Kim, J. S.; Chris, R.; Taitt, C. R.; Ligler, F. S.; Anderson, G. P. Sens. Instrum. Food Qual. 2010, 4, 73–81.

Figure 3. Circular dichroism spectra of sdAb B4 (30 µg/mL). Initial spectrum at 25 °C (blue), spectrum at 85 °C (red), and spectrum after sdAb had been cooled to 25 °C (green). Shows recovery of secondary structure upon cooling.

minimizing this factor. B4 immobilized to a microsphere and used in fluid array immunoassays failed to give similar results; very poor detection of ricin was achieved (Figure S3, Supporting Information). Either the covalent attachment of B4 to the microsphere interferes with it is binding site and/or the high off rate is more detrimental in this format. Additional work examined the thermal stability of B4 by monitoring its secondary structure by CD while being heated and cooled through multiple cycles. Figure 3 shows the CD spectrum of B4 at room temperature, heated to 85 C, and then again at room temperature. This indicates that the bulk of B4’s secondary structure is lost upon heating but that B4 refolds substantially upon cooling. The refolding of sdAb after heat denaturation has been previously observed and reported.23-25 To better investigate this phenomena, B4 was thermally cycled multiple times with the ellipticity monitored at 205 nm (Figure 4a). This experiment showed that the loss in B4’s secondary structure upon thermal cycling is most substantial during the first cycle, with only minimal loss occurring in subsequent cycles. We had seen a similar result with C8. To evaluate how the changes in secondary structure correlated to activity, samples of B4 and C8 which had been thermally cycled were tested for their ability to bind to ricin, monitored by SPR (Figure 4b). B4 was found to lose ∼25% of its activity over four cycles, while C8 was found to be a bit more thermally stable, losing only ∼10% of its initial activity. These tests were conducted at concentrations sufficiently low so that the measurement would be sensitive to relatively small changes in activity, whereas earlier experiments required substantial degradation to occur before an effect would be observed.14 (23) Perez, J. M. J.; Renisio, J. G.; Prompers, J. J.; van Platerink, C. J.; Cambillau, C.; Darbon, H.; Frenken, L. G. J. Biochemistry 2001, 40, 74–83. (24) Dumoulin, M.; Conrath, K.; Van Meirhaeghe, A.; Meersman, F.; Heremans, K.; Frenken, L. G. J.; Muyldermans, S.; Wyns, L.; Matagne, A. Protein Sci. 2002, 11, 500–515. (25) Goldman, E. R.; Anderson, G. P.; Conway, J.; Sherwood, L. J.; Fech, M.; Vo, B.; Liu, J. L.; Hayhurst, A. Anal. Chem. 2008, 80, 8583–8591.

Figure 4. Thermalstability of sdAb B4. (A) Heating and cooling cycles of clone B4 monitored by circular dichroism spectroscopy. To evaluate the manner in which the sdAb unfolded and refolded, sdAb B4 was heated to 85 °C and then cooled at varying rates while the ellipticity at 205 nm was monitored. Results show that sdAb refold as rapidly as they unfold. (B) Measurement of binding activity following heating to 85 °C. The SPR signal (resonance units) generated by the addition of 10 nM sdAb to a surface coated with ricin was measured following the indicated number of heating cycles.

CONCLUSIONS Along with reevaluating the kinetic parameters of a number of antiricin sdAb, this work identified and evaluated an antiricin B chain sdAb. The use of this B chain binding sdAb along with an A chain binding sdAb provided sensitive detection of ricin (