Competitive, immunometric assay for fusion protein quantification: Protein production prioritization

Analytical Biochemistry 446 (2014) 1–8

Contents lists available at ScienceDirect

Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio

Competitive, immunometric assay for fusion protein quantification: Protein production prioritization Hyun-Hee Cho a, Edward Alderman b,c,⇑, Natasha Kreder a, Roxana Garcia Caro a,b, Kristen Leong b, Michael F. Miller b, W. Adam G. Hill a, Pramod Pandey a,⇑ a b c

Novartis Institutes for BioMedical Research, Inc., Cambridge, MA 02139, USA Bioscale, Inc., Lexington, MA 02421,USA Immunologic Consulting, LLC, Framingham, MA 01702, USA

a r t i c l e

i n f o

Article history: Received 27 June 2013 Received in revised form 26 September 2013 Accepted 30 September 2013 Available online 10 October 2013 Keywords: Purification Fusion protein Affinity tag Competitive assay

a b s t r a c t Effective drug discovery demands the availability of microgram to gram quantities of high-quality protein encoded by novel transcripts. Protein expression vectors designed for large-scale protein production often include one or more specific tags to such transcripts, to simplify the purification of the targeted protein. Optimization of the complex expression and purification process requires the evaluation of multiple expression candidate clones to identify a production-suitable construct in terms of quality and final protein yield. Efficiency of the entire expression screening process is typically assessed by direct visualization of the banding patterns from whole-cell lysates on SDS–PAGE gels, by direct staining and/or immunoblotting, using antibodies against the tag or the protein of interest. These techniques, generally run under denaturing conditions, have proven to be only marginally predictive of the purification yield and authentic folding for native proteins. Small-scale, multiparallel affinity purification followed by SDS–PAGE analysis is more predictive for expression screening; however, this approach is labor intensive and time consuming. Here we describe the development of an alternative expression efficiency assessment technique, designed to evaluate the accessibility of affinity tags expressed with the desired fusion proteins, using acoustic membrane microparticle assay technology on the ViBE protein analysis workstation. Ó 2013 Elsevier Inc. All rights reserved.

Taking full advantage of high-throughput genomic studies, which have elucidated novel sequences related to disease states or to cellular responses to proposed activity modulators, demands increased throughput in the expression and purification of the proteins encoded by such transcripts. A common approach to fulfilling this need has been the generation of chimeric fusion proteins using specific expression vectors [1–4]. The design of such expression vectors has evolved to increase target protein solubility [5], to promote appropriate target protein folding [6,7], and to simplify the target protein purification processes [2,4]. Unfortunately, the selection of appropriate clones to advance for larger scale production, typically made on the basis of relative band strength on stained SDS–PAGE gels and/or Western immunoblots, has proved to be problematic. Neither gels nor immunoblots are representative of the native state of the desired protein while in solution, nor do they provide quantitative results [8]. Consequently, multiple constructs must be tested in parallel to ensure the generation ⇑ Corresponding authors. Fax: +1 (781) 430 6801 (E. Alderman). Fax: +1 (617) 871 4088 (P. Pandey). E-mail addresses: [email protected] (E. Alderman), pramod.pandey@ novartis.com (P. Pandey). 0003-2697/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ab.2013.09.031

of the desired protein, generally incurring significant expense and unanticipated delays in downstream discovery research and process development. One of the most widely used fusion sequences is the polyhistidine affinity tag, which encodes a series of four or more histidine residues within the targeted sequence or on its N- or C-terminus. Upon expression, affinity purification of the polyhistidine structure, typically using chelated metal (Ni2+, Co2+, or Zn2+) affinity chromatography, permits copurification of the attached, targeted protein. However, occlusion, constriction, or other obstruction of the histidine-rich portion of the expressed fusion protein hinders its binding to the affinity matrix, thus compromising the efficiency of the purification [9]. Here we describe the development of a novel technique to provide a rank order of prospective expression clones for purification, using histidine-enriched sequences as an example, on a variety of fusion proteins derived from Escherichia coli and insect cell expression systems. This method measures the presence and accessibility of polyhistidine-enriched fusion proteins contained in lysates prepared from such protein production candidate lines, by their ability to competitively inhibit the binding of a tracer

2

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8

molecule (fluorescein-labeled monoclonal anti-pentahistidine antibodies) to paramagnetic bead-immobilized protein with an accessible polyhistidine tag (Fig. 1). Acoustic membrane microparticle (AMMP) assay technology combines the specificity of immunological antigen capture techniques with the sensitivity of microelectromechanical systems (MEMS) sensors, using magnetic microparticles to create novel, nonoptical, detection procedures. For this proof-of-concept study, we used 12 target genes encoding proteins ranging in size from 27 through 235 kDa. Six of these proteins were expressed in E. coli and six were expressed in insect (Sf9) cells using a baculovirus expression system. AMMP assays were performed using crude lysates and the resulting data were compared with the actual purification yields following Ni–NTA column purification. Materials and methods Principles of the AMMP assay The AMMP assay has been described previously [10]. In this particular instance, the technique measures the ability of an unknown fusion protein, putatively containing an N-terminal, C-terminal, or internal sequence of 4–10 histidine residues, to competitively inhibit the binding of fluorescein-labeled anti-pentahistidine antibodies (anti-5His) to a similarly expressed fusion protein known to possess an accessible C-terminal hexahistidine sequence, which has been immobilized onto the surface of paramagnetic beads. Signal is generated in the system when the fluorescein-labeled tracer antibody ‘‘bridges’’ the hexahistidinecoupled beads to the anti-fluorescein-coupled sensor surface—the vibrational frequency of the sensor membrane decreases in direct proportion to the number of beads bound to the sensor surface (Fig. 1). Nonspecifically bound materials are washed from the sensor surface using a flow of running buffer (0.02 M phosphate buffer, containing 0.15 M NaCl and 1% (v/v) Tween 20). Between measurements, the sensor surface is regenerated using a similar flow of BioScale regeneration solution (P/N: 75018-0005), followed by running buffer to prepare for analysis of the next series of samples. Bead preparation Solutions containing 10, 20, and 40 lg (0.5, 1.0, and 2.0 nmol, respectively) of a 20-kDa recombinant protein including a C-termi-

nal hexahistidine sequence (His target) are coupled to 1-mg aliquots (33 ll) of paramagnetic microbeads according to the manufacturer’s recommendations. The resulting mixtures are incubated with continual end-over-end mixing, for 18 h at 25 °C, then washed three times with phosphate-buffered saline (PBS) containing 0.05% Tween 20, and resuspended to a final volume of 33 ll in PBS containing 1% bovine serum albumin, for storage at 4 °C prior to use. Similar microbead preparations are made using 20 lg bovine serum albumin (BAH65; Equitech-Bio) per milligram of beads for use as the ‘‘irrelevant bead’’ control. These reagents are available from BioScale (P/N: 75081-0003). Tracer preparation An anti-pentahistidine antibody (34440; Qiagen) was conjugated with fluorescein according to the manufacturer’s recommendations (50541; BioScale). The resulting tracer antibody preparation was then stored in 0.02 M phosphate-buffered saline containing 0.05% bovine serum albumin at 4 °C for later use. ‘‘Irrelevant antibody’’ control samples were prepared as above, using rat anti-human IL-6 monoclonal antibody (75062-0003; BioScale). Standard preparation Standards (500 nM) were prepared by the addition of 100 lg of the His target fusion protein (200 ll  500 lg/ml) to 9.8 ml E. coli or Sf9 ‘‘null’’ lysate preparations. Resulting standard preparations were stored at 80 °C as 1-ml aliquots for later use. AMMP assay protocol For analysis, all samples and standards were serially diluted (threefold, eight places) into sample dilution buffer (20 mM sodium phosphate buffer containing 450 mM NaCl and 0.05% Tween 20). Three columns of a standard 96-well microplate were used for each dilution series—80 ll sample was dispensed in duplicate to columns 2 and 3 while column 1 contained only diluent. All three columns received His target protein-coated beads (1.5  105 beads/well). The ViBE workstation was programmed to deliver additional reagents at time-controlled intervals to ensure appropriate incubation periods. After 30 min of shaking preincubation, the ViBE workstation was programmed to deliver tracer antibody to each column in the sample set at regular intervals, thus permitting 60 min incubation for each well. Following the incubation, aliquots from each sample column (8 wells/column) were injected for analysis. Additionally, irrelevant bead (20 ll immobilized bovine serum albumin (BSA)–beads, 1.5  105/well) and irrelevant tracer (20 ll fluorescein-labeled rat anti-IL-6 monoclonal antibody, 200 ng/ml) controls were included periodically to ensure assay performance and assess freedom from host-cell-attributable nonspecific assay interference. Cell culture and protein expression

Fig.1. Schematic representation of AMMP competitive immunometric assay. Red squares represent His-tagged standard ‘‘target’’ protein; blue circles represent multi-histidine tag; yellow octagons represent unknown, His-tagged protein; green diamonds represent fluorescein label on (black) anti-histidine ‘‘tracer’’ antibody. Immobilized (gray) antibody on sensor surface represents anti-fluorescein.

A total of 12 cultures of E. coli or Sf9 cells, each encoded with a different human hexahistidine affinity tag fusion protein, were used in this study (Table 1). Six E. coli proteins (P1–P6) were cultured in 150 ml cultured TB medium at 37 °C and induced at 18 °C with 0.25 mM isopropyl-b-D-1-thiogalactopyranoside for 16 h. Exponentially growing Sf9 cells were cultured in 150 ml culture vessels, at a density of 1.2  108 cells/ml. After 24 h incubation at 27 °C, the cells were infected (1:50, v/v) with six different viral constructs. Infected cells were incubated for 72 h at 27 °C, shaking at 100 rpm. Cell viabilities and diameters were measured with a Vi-CELL XR analyzer (trypan blue exclusion method; Beckman

3

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8 Table 1 Summary of results for the 12 proteins in the study. Protein code

Expression system MW (kDa) His-tag position Actual yield (lg/ml) ID50 ID50 ratio AMMP rank Normalized predicted yield Normalized actual yield

Std

P1

P2

P3

P4

P5

P6

P7

P8

P9

P10

P11

P12

E. coli 59.3 N-term 184 0.003 3.663 1 65.2 5.58

E. coli 50.1 N-term 0.5 No fit No fit NC No fit 0.01

E. coli 27.4 N-term ND No fit No fit NC No fit ND

E. coli 31.7 C-term 33 0.109 0.105 2 1.0 1.0

E. coli 35.6 N-term 9 0.505 0.023 3 0.25 0.27

E. coli 35.6 N-term 9 0.480 0.024 3 0.26 0.27

Sf9 98.6 C-term 10 1.086 0.011 5 0.42 0.33

Sf9 121 C-term 67 0.047 0.242 2 11.4 2.34

Sf9 235 N-term ND No fit No fit NC No fit ND

Sf9 48.6 C-term 14 0.686 0.017 4 0.32 0.48

Sf9 37.1 N-term 156 0.045 0.257 1 3.72 5.40

Sf9 35.6 N-term 29 0.159 0.072 3 1.0 1.0

E. coli 20.1 C-term – 0.0115 1 – – –

ID50, 50% inhibitory dilution; NC= Not Calculated; ND= Not Detected

Coulter) and cells were harvested when viabilities were between 70 and 80%.

Sample preparation and characterization for AMMP assays E. coli lysates were prepared in lysis buffer containing 50 mM Tris (pH 7.5), 500 mM NaCl, 10% glycerol, 1 mM tris(2-carboxyethyl)phosphine (TCEP), 0.05% Triton X-100, benzonase (10 units/ml), EDTA-free protease inhibitor, and lysozyme (1 mg/ml). In addition to the six test sample preparations (P1– P6), a nontransfected, null lysate was prepared. Insect (Sf9) cell lysates (P7–P12), as well as an uninfected null lysate, were prepared in the same buffer as E. coli lysates, omitting the lysozyme. Lysates were centrifuged at 14,000g for 30 min and soluble (in supernatant) expression products were evaluated by SDS–PAGE (Fig. 2) and Western blots using anti-pentahistidine antibody (Fig. 3). Soluble proteins present in each lysate were used for AMMP assay expression testing. Total protein concentration for each lysate was determined by microBCA analysis and several 10 lg/ml aliquots of each lysate were stored at 20 °C for later use.

Sample purification The cleared lysates (2 ml of lysis buffer added to 50 ml of frozen and thawed pellet) from the E. coli and Sf9 cells were incubated with 50 ll of preequilibrated Ni–NTA Superflow resin (Qiagen, Cat. No. 1018401) for 30 min at 4 °C with gentle rocking (The QIAexpressionist handbook). The resin was batch-washed three times with 10 volumes of wash buffer (50 mM Tris, pH 7.5, 300 mM NaCl, 0.05% Triton X-100, 1 mM TCEP, and 20 mM imidazole) and then transferred to 1 ml columns. The proteins were eluted from the columns with 200 ll of 200 mM imidazole in the wash buffer. The eluates were dialyzed with a 10 K molecular weight cutoff membrane (Thermo Scientific, 87730) for 2 h against 500 ml of buffer containing 50 mM Tris, pH 7.5, 150 mM NaCl, and 20% glycerol and then overnight with a new 500 ml of buffer. The purified proteins were run on SDS–PAGE and stained with Coomassie blue for visualization (Fig. 4). Protein concentration was determined using the Bradford method and the final concentrations were normalized based on the purity of each sample (Table 1). Purity assessment of the protein bands (Fig. 4) was done using the ImageJ software [11].

Fig.2. Coomassie blue-stained SDS–PAGE gel for E. coli and insect cell soluble lysates. Lanes 1–6, six different His-tagged proteins (P1–P6) expressed in E. coli; lanes 7–12, six different His-tagged proteins (P7–P12) expressed in Sf9 (insect) cells using a baculovirus expression system. The arrows indicate the bands corresponding to candidate proteins that had the highest yields.

4

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8

Fig.3. Western blot analysis of samples 1–12 using anti-pentahistidine antibody. Lanes 1–6, six different His-tagged proteins (P1–P6) expressed in E. coli; lanes 7–12, six different His-tagged proteins (P7–P12) expressed in Sf9 (insect) cells using baculovirus.

Fig.4. Postpurification SDS–PAGE gel of samples 1–12 final product. Lanes 1–5 are samples from proteins purified from E. coli soluble lysates (P1–P5). Sample 6 was not run in this gel because sample 5 and sample 6 were from the same lysate preparation. Lanes 7–12 contain purified His-tagged proteins from Sf9 (insect) cells using the baculovirus expression system.

Results Reagent concentration optimization Optimal bead-loading concentration for the His target protein and appropriate tracer (fluorescein-labeled anti-5His) antibody concentrations were determined in ‘‘checkerboard’’ assays (Figs. 5 and 6). The beads used in this experiment (irrelevant bead control

(2.0 nmol BSA/mg bead) and His target beads coated at 0.5, 1.0, or 2.0 nmol/mg bead) were incubated with E. coli null lysate containing 500 nM His-tagged standard serially diluted threefold through eight places (S1–S8). After 30 min shaking incubation at room temperature, 20-ll aliquots of irrelevant tracer (fluorescein-labeled anti-IL-6 at 2.0 nM) or relevant tracer (fluorescein-labeled anti5His antibody) at 6.67, 3.33, 1.67, 0.83, or 0.42 nM were added as indicated. Shaking incubation at room temperature was continued

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8

5

Fig.5. Checkerboard analysis used to select appropriate tracer antibody concentration. The condition selected, a 3.33 nM solution of anti-5His tracer antibody, provides maximal signal strength with minimal signal attributable to nonspecific binding due to irrelevant protein-labeled beads or irrelevant antibody binding in the system.

Target Density Titraon

0.9000 0.7000 0.6000 0.5000 0.4000 0.3000 0.2000 0.1000

AMMP (Ru)

0.8000

2.00 1.00 0.50 irrev tracer

2.00

1.00

0.50

irrev tracer

0.0000

Fig.6. Checkerboard analysis used to select appropriate target density (concentration of the His-tagged target bound to magnetic bead surface). Loading 1 nmol of protein/mg beads resulted in signal strength comparable to the 2-nmol load and nearly double the 0.5-nmol load, while showing nonspecific binding levels lower than the 2-nmol load and indistinguishable from the 0.5-nmol load.

60 min prior to automatic injection into the ViBE analyzer. In general, the greater sensitivity gained by decreasing the concentration of His target immobilized to bead surfaces (lower 50% inhibitory concentration (IC50) value) was somewhat offset by the diminished signal range observed and, thus, loss of precision in the assay. Increased tracer antibody concentration could restore much of the signal range, but the effect was ultimately limited by increases in nonspecific binding, as evidenced by increased signal generated in the presence of irrelevant antibody. Consequently, optimal concentrations of the reagents for this assay were selected on the basis of (1) maximum S1 signal, (2) minimum S8 signal, and (3) minimum signal attributable to irrelevant bead or irrelevant tracer. By these criteria, a bead-loading density of 1 nmol of His target per milligram of beads (Fig. 6) and 3.33 nM anti-5His tracer antibody (Fig. 5) were selected for testing samples. Sample testing The signal obtained for each of the serially diluted standards, when plotted against the known concentrations of these standards, returns the IC50 (6.04 nM, Figs. 7 and 8) based on a four-parameter

logistic analysis [12–14]. Discrete standard concentrations were back-calculated from this standard logistic equation and compared with the ‘‘nominal’’ concentrations to calculate recovery of the standard. An acceptable range of ‘‘raw signal’’ values reflecting standard concentration recovery of 100 ± 20% was established from these calculations and applied to the raw signal values obtained for each sample dilution to determine valid results. Consequently, back-calculated sample concentrations were determined for data points falling within the acceptable raw signal range defined by standard recovery. Similarly, such analysis of the same standard signal data plotted against the known serial dilution factors returns the 50% inhibitory dilution (ID50), thus relating the dilution data to the concentration data. Consequently, the ID50 ratio, defined as ID50standard/ID50sample, can be used as a measure of the potency of each sample relative to the standard (Table 1). The relative potency of a sample will depend on the concentration as well as the accessibility of the Histagged fused to the protein of interest. The actual purification yield achieved for a given sample will depend similarly on these two factors, as well as the overall efficiency of the purification process used. As such, the relative

6

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8

Fig.7. AMMP data obtained from the ViBE workstation for E. coli samples. The IC50 of the His-tagged standard was determined (in the inset graph) to be 6.04 nM. Ratios of the ID50 values for samples P1–P6 to the ID50 observed for the His-tagged standard, calculated from fitting the data points to four-parameter logistic curves, reveal the potency of each sample relative to the standard. Thus, the rank order (molar) of P1 > P4 > P5  P6 > P2  P3 is observed. Data points reflect the mean of nine independent determinations, and error bars represent the standard deviations.

Fig.8. AMMP data obtained from the ViBE workstation for insect (Sf9) cell lysates. The IC50 of the His-tagged standard was determined (in the inset graph) to be 6.04 nM. Ratios of the ID50 values for samples P7–P12 relative to the ID50 observed for the His-tagged standard, calculated from fitting the data points to four-parameter logistic curves, reveal the potency of each sample relative to the standard. The resulting rank order (molar) for the Sf9 samples is P11 > P8 > P12 > P10 > P7 > P9. Data points reflect the mean of nine independent determinations, and error bars represent the standard deviations.

potency from the AMMP assay can be used to rank the expected purification yields. As a step beyond basic rank order, estimates of potential purification yields can made by assuming the accessibility of the His tag is equivalent for the samples and the standard. Since the actual yields are expressed on a mass basis, the predicted yields were calculated on a mass basis as well. To simplify the com-

parison both the predicted and the actual mass-based yields were normalized using one of samples from each set—P4 for the E. coli samples and P12 for the Sf9 samples (Table 1). Each sample was tested a total of nine times—3 days  3 analysts, with results as indicated in Figs. 7 and 8 and summarized in Table 1 and Supplementary Table S1. The ID50 ratios determined

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8

from the raw signal data plotted vs the logarithm of the dilution for each 10 lg/ml sample were used to rank the expected purification yields of the samples (Table 1). For the E. coli samples, the rank order was P1 > P4 > P5  P6 > P2  P3 (Fig. 7). Sample P1 was roughly 3.6-fold more potent than the standard, while sample P4 was roughly 10-fold less potent. Samples P5 and P6 were nearly 50-fold less potent than the standard, and samples P2 and P3 reflected concentrations too low to quantify. Samples P5 and P6 were generated from the same lysate; thus the samples showed similar results in the AMMP assay as expected. For the E. coli lysates, all samples other than P1 looked similar in gels run with soluble protein (Fig. 2). As a result, based on the soluble protein gel alone it would be difficult to select appropriate cell lines for large-scale production. For the Sf9 samples, neither the soluble lysate gel nor the Western blots were predictive of the rank order of the actual yields after purification. Samples P7–P12 gave similar banding patterns in the soluble lysate gel (Fig. 2), while the rank order (P11 > P10 > P12 > P7 = P8 > P9) obtained from Western blot analysis (Fig. 3) varied significantly from the AMMP assay prediction and the actual protein yields (P11 > P8 > P12 > P10 > P7 > P9). The AMMP assay results indicate that samples P8 and P11 were roughly 4-fold less potent than the standard, while sample P12 was nearly 14-fold less potent. Samples P10 and P7 scored about 60- and 90-fold less potent than the standard, and sample P9 was below detection and quantitation limits. Interestingly, protein sample P8, which ranked very low by Western blot analysis, scored high in both the AMMP assay and the purified product SDS–PAGE gel (Fig. 4). Also, samples P8 and P11 have similar measured potencies; however, because P8 has a significantly higher molecular weight it has a higher predicted purification yield when expressed on a mass basis. As the results in Table 1 show, rank ordering based on ID50 ratio is effective for prioritizing candidate proteins based on the potential for purification—the candidates with the highest actual yields were ranked at the top of each group by the AMMP assay. Comparing the normalized predicted and actual yields for the E. coli group, the AMMP assay provides accurate estimates (within 10%) of the relative yields for the lower-producing candidates but significantly overestimates the observed yield for the top-producing candidate, P1. For the Sf9 group, the AMMP assay provides relative yield estimates within 33% of the observed yields for all the candidates except for P8. These discrepancies may be related to the impact of the His-tag availability or the molecular weight of the target on the purification process being distinct from the assay response.

Discussion In this postgenomic era of biopharmaceutical discovery research, there is an ever increasing demand for more rapid delivery of modestly characterized proteins in ever larger quantities. Largescale expression of recombinant proteins in E. coli or baculovirustransfected insect cells utilizes complex vectors containing not only selective elements, but also chimeric sequences such as hexahistidine or glutathione S-transferase coding regions, to simplify the purification of the resulting fusion proteins. The His-tag sequence is particularly popular in these expression laboratories. It can be fused internally or to either the N- or the C-terminus of the targeted protein and is sufficiently small (0.8 kDa) that the sequence rarely requires cleavage. Efficient production of the fusion protein of interest requires the timely selection of a single host production line from several candidate lines, based not only on the stable generation of protein, but also on the accessibility of the chimeric affinity tag used to facilitate its purification. Although SDS–PAGE and Western immuno-

7

blots give reasonable insight into the protein generation parameters in terms of overall expression yield, they are not as predictive of tag accessibility that ultimately facilitates the purification of the tagged protein. This often leads to the selection of cell lines that produce copious amounts of protein that cannot be purified using standard affinity-chromatographic techniques. Similarly, a more recent optical interferometric method has been reported to provide rapid, predictive expression results among closely related protein targets, but with limited utility when applied to protein targets with widely variant molecular weights [15]. Here, we have developed a technique to rapidly and efficiently prioritize candidates based on expected purification yield, making use of BioScale’s nonoptical AMMP assay technology run on an automated ViBE workstation. This system measures minute changes in the resonant frequency of MEMS sensors in response to the binding of microparticles [10]. Consequently, when the tracer antibody is ‘‘sandwiched’’ between the bead-immobilized His target protein and the sensor-immobilized anti-fluorescein antibody, the resonance frequency of the sensor is lowered in direct proportion to the number of captured beads. Thus, in the absence of any poly-His-containing inhibitor, the fluorescein-labeled tracer antibody forms a complex with the bead-immobilized His–target fusion protein and the sensor surface, leading to a stable, positive signal in the assay. This signal is diminished in direct proportion to the concentration of accessible poly-His moieties contained within each sample. The immunometric format chosen for this assay and the similarity of the standards (spiked into null lysates) to the test samples ensure that the resulting response curves are parallel, thereby simplifying analysis of sample potency relative to that of the standards. For these experiments, we opted for conditions to yield nano- to picomolar sensitivity for His-tagged fusion proteins, namely 20 lg (1 nmol) of His-tagged protein per milligram of beads as target and 500 ng/ml (3.33 nM) antibody tracer. These concentrations resulted in the delivery of approximately 250 fmol of immobilized target and approximately 325 fmol of tracer antibody per well. Assay sensitivity is dependent upon the concentrations of both the immobilized target and the tracer antibody in a given assay well. Decreased immobilized target density generally results in diminished background and is the major determinant of the lower limit of detection achieved in the assay. The major component in determining assay precision is the concentration of tracer antibody used in the system. As tracer antibody concentrations decrease, both signal range and Hill slope are reduced, resulting in ‘‘flatter’’ curves, while resulting IC50 values are not significantly altered. For this type of assay, the optimal tracer antibody concentration (2 nM in this study) is that which roughly approximates the total target concentration in a given well, providing maximal signal range while maintaining minimal nonspecific binding. The standard solution returned similar IC50 values regardless of the concentration of immobilized target (0.5–2.0 nM) or tracer antibody (0.42–6.67 nM) used in the assay, demonstrating the robust nature of the assay. The assay provides a rapid, convenient method of ranking candidate proteins based on the potential to be purified with good yield. For the assay to provide a response, the His tag must be accessible. Thus, it provides a unique test regarding the purification potential compared to SDS–PAGE or Western blot analyses, which rely on denaturing conditions. Additionally, results are obtained with much less effort and time than is required by these alternative methods or for direct testing via small-scale purification. The current results support the use of the AMMP assay for rank ordering to prioritize candidate proteins for purification. Quantitative interpretation of results for predicting ultimate purification yields may be influenced by differing effects of His-tag availability on the purification process versus the assay, the

8

Competitive immunoassay for fusion proteins / H.-H. Cho et al. / Anal. Biochem. 446 (2014) 1–8

molecular weight of the candidate protein, the potential of the candidate protein to form multimers and/or aggregates upon purification, and the overall efficiency of the purification process ultimately employed. Additionally, the predicted yields are reflective only of the metal-affinity chromatographic steps and will not estimate the efficacy of peak-pooling or peak-shaving of fractions separated by ion-exchange, molecular-sieving, or other chromatographic techniques applied following the affinity purification step(s). The reproducibility, sensitivity, and precision of this assay indicate the potential utility of immunometric AMMP assays for the detection and quantification of additional analytes, which may present as ‘‘proforms,’’ including a number of enzymatically active proteins and many of the smaller neuropeptides that are generally difficult to quantify. One such class of small peptide hormones includes oxytocin and vasopressin, each of which is only 9 amino acids in length, but generally associated with chaperone or carrier proteins. It is envisioned that a similarly formatted competitive, immunometric, AMMP assay using the intact proform or chaperone–analyte complex as immobilized target may provide rapid and precise quantitation for these historically ‘‘difficult’’ analytes. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ab.2013.09.031. References [1] I. Hunt, From gene to protein: a review of new and enabling technologies for multi-parallel protein expression, Protein Expression Purif. 40 (2005) 1–22.

[2] D. Walls, S.T. Loughran, Tagging recombinant proteins to enhance solubility and aid purification, Methods Mol. Biol. 681 (2011) 151–175. [3] M.E. Kimple, J. Sondek, Overview of affinity tags for protein purification, Curr. Protoc. Protein Sci. 9 (2004) 9.1–9.9.19. [4] C. Young, Z.T. Britton, A.S. Robinson, Recombinant protein expression and purification: a comprehensive review of affinity tags and microbial applications, Biotechnol. J. 7 (2012) 620–634. [5] D. Waugh, Making the most of affinity tags, Trends Biotechnol. 23 (2005) 316– 320. [6] S. Nallamsetty, D. Waugh, Solubility-enhancing proteins MBP and NusA play a passive role in the folding of their fusion partners, Protein Expression Purif. 45 (2006) 175–182. [7] I. Choi, K. Han, C. Kim, K. Ryu, B. Kim, K. Kim, S. Kim, T. Kang, H. Shin, K. Lim, H. Kim, J. Hyun, B. Seong, Protein solubility and folding enhancement by interaction RNA, PLoS ONE 3 (2008). [8] H. Nguyen, B. Martinex, N. Oganesyan, R. Kim, An automated small-scale protein expression and purification screening provides beneficial information for protein production, J. Struct. Funct. Genomics 5 (2004) 23–27. [9] E. Ueda, P. Gout, L. Morganti, Current and prospective applications of metal ion-protein binding, J. Chromatogr. A 988 (2003) 1–23. [10] Y. Zhong-Hua, L. Madison, A. Burkhardt, J. Yu, O. Tayber, Z. Li, D. Wu, H. Loke, D. Wysong, N. Iartchouk, M. Smith, M. Kuranda, P. Li, M. Bembenek, Analysis of two pharmacodynamic biomarkers using acoustic micro magnetic particles on the ViBE bioanalyzer, Anal. Biochem. 410 (2010) 13–18. [11] C.A. Schneider, W.S. Rasband, K.W. Eliceiri, NIH Image to ImageJ: 25 years of image analysis, Nat. Methods 9 (2012) 671–675. [12] M. Haven, P. Orsulak, L. Arnold, G. Crowley, Data-reduction methods for immunoradiometric assays of thyrotropin compared, Clin. Chem. 33 (1987) 1207–1210. [13] R. Dudley, E. Edwards, R. Ekins, D. Finney, I. McKenzie, G. Raab, D. Rodbard, R. Rodgers, Guidelines for immunoassay data processing, Clin. Chem. 31 (1985) 1264–1271. [14] D. Rodbard, G. Fraier, Statistical analysis of radioligand assay data, Methods Enzymol. 37 (1975) 3–22. [15] J. Tian-Yu, S. Licht, G. Pardee, A. Bhat, Y. Cao, W. Gao, E. Sangalang, I. Zaror, Binding rate screen—a high-throughput assay in soluble lysate for prioritizing protein expression constructs, Anal. Biochem. 399 (2010) 276–283.