NanoLuc Complementation Reporter Optimized for Accurate

Nov 16, 2015 - E-mail: [email protected]. ACS AuthorChoice - This is an open access article published under an ACS AuthorChoice .... Applicatio...
0 downloads 10 Views 3MB Size
This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Articles pubs.acs.org/acschemicalbiology

NanoLuc Complementation Reporter Optimized for Accurate Measurement of Protein Interactions in Cells Andrew S. Dixon,†,§ Marie K. Schwinn,† Mary P. Hall,† Kris Zimmerman,† Paul Otto,† Thomas H. Lubben,† Braeden L. Butler,† Brock F. Binkowski,† Thomas Machleidt,† Thomas A. Kirkland,‡ Monika G. Wood,† Christopher T. Eggers,† Lance P. Encell,*,† and Keith V. Wood† †

Promega Corporation, Madison, Wisconsin 53711, United States Promega Biosciences Incorporated, San Luis Obispo, California 93401, United States



S Supporting Information *

ABSTRACT: Protein-fragment complementation assays (PCAs) are widely used for investigating protein interactions. However, the fragments used are structurally compromised and have not been optimized nor thoroughly characterized for accurately assessing these interactions. We took advantage of the small size and bright luminescence of NanoLuc to engineer a new complementation reporter (NanoBiT). By design, the NanoBiT subunits (i.e., 1.3 kDa peptide, 18 kDa polypeptide) weakly associate so that their assembly into a luminescent complex is dictated by the interaction characteristics of the target proteins onto which they are appended. To ascertain their general suitability for measuring interaction affinities and kinetics, we determined that their intrinsic affinity (KD = 190 μM) and association constants (kon = 500 M−1 s−1, koff = 0.2 s−1) are outside of the ranges typical for protein interactions. The accuracy of NanoBiT was verified under defined biochemical conditions using the previously characterized interaction between SME-1 β-lactamase and a set of inhibitor binding proteins. In cells, NanoBiT fusions to FRB/FKBP produced luminescence consistent with the linear characteristics of NanoLuc. Response dynamics, evaluated using both protein kinase A and β-arrestin-2, were rapid, reversible, and robust to temperature (21−37 °C). Finally, NanoBiT provided a means to measure pharmacology of kinase inhibitors known to induce the interaction between BRAF and CRAF. Our results demonstrate that the intrinsic properties of NanoBiT allow accurate representation of protein interactions and that the reporter responds reliably and dynamically in cells.

P

(116 kDa monomer) has also been used as a split reporter,4 but its size and requirement for oligomerization further impose interaction relationships that may be inconsistent with the target proteins. Furthermore, reporters such as β-galactosidase that rely on the accumulation of an enzymatic product for detection are hindered in their ability to reveal process dynamics. Assays based on split luciferases are often preferred due to their sensitivity and simplicity. Luminescence provides an instantaneous indication of reporter activity in living cells and is readily generated by adding substrate to the growth medium. However, although commonly regarded for producing quantitative data, split luciferases have been evaluated largely for assay precision rather than accuracy.5−8 While reproducibility and correlation with known behaviors of interacting proteins have been shown, quantitative concordance of assay data with the underlying molecular interaction has been difficult to substantiate. Hence, reinforced by the known limitations associated with many split reporters, doubts persist on the general suitability of

roteins are central to the physiological dynamics of living cells, forming interactive networks that are responsive to evolving circumstances (e.g., environmental conditions, stress or pathological situations, or the action of drugs).1 The interactions arise through a complex interplay of structural conformations, relationships with other molecules, and microenvironmental surroundings. Due to the complexity of these functionally contingent ensembles, analytical methods are needed that do not disturb the native cellular context. Split reporters have been commonly used for this purpose, where complementary reporter fragments are genetically fused onto a pair of interacting proteins.2 The fragments normally exhibit little reporter activity separately, but regain activity when brought together through the interaction of their fused partners. Although conceptually straightforward, the ability of split reporters to provide quantitative data without perturbing the underlying interaction dynamics is not assured. Because association of the reporter fragments is required for the reconstituted activity, the intrinsic binding affinity of the fragments could bias the behavior of their fusion partners. Split fluorescent proteins, for example, do not reversibly associate, and chromophore maturation tends to be slow, thus obscuring the dynamics of the target proteins under study.3 β-Galactosidase © 2015 American Chemical Society

Received: September 19, 2015 Accepted: November 16, 2015 Published: November 16, 2015 400

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology

Figure 1. Nluc dissection and optimization. (a) Nluc topology model9 consisting of a 10-stranded beta-barrel. The identified dissection point (red arrow) occurs between residues 156 and 157 and generates a 156-amino acid polypeptide (156) and a 13-amino acid peptide (native peptide, NP). (b) Nluc Western blot of HEK293T lysates expressing empty vector (−), 156, 11S, or Nluc. β-actin was included as a loading control. (c) Intracellular halflife of Nluc, 156, and 11S in HeLa cells following treatment with cycloheximide, as measured by enzyme activity, i.e., normalized relative luminescence units (RLU; with saturating NP for 156 and 11S). (d) Thermal stability of purified Nluc, 156, and 11S, based on remaining activity at ambient temperature in the presence of saturating NP added 30 min after temperature challenge. (e) Specific activities of 156 and 11S normalized to Nluc. Equimolar amounts of each were used to determine Vmax values (under conditions of saturating NP for 156 and 11S). (f) Titration of 11S with various peptides demonstrating the broad range (>105) of binding affinity. KD values (shown in parentheses) were determined by fitting the data to a one site specific binding model. For panels c−f: n = 3, variability displayed as SD.

designed to be quantifiable within living cells at relevant concentrations and temperatures. We accomplished this through the application of two unique design strategies. First, beyond simply identifying an optimal split site, we further enhanced efficacy through structural optimization of the individual components. Although this approach had already been applied in the development of Nluc,9 we expanded it to address issues specific to a dynamic binary structure. The second strategy resulted from our fortuitous discovery that Nluc could be dissected 13 amino acids from its C-terminus. We envisioned that such a small tag would minimize steric conflicts on fusion partners where geometric constraints may be critical. This design strategy also allowed reporter stability and intrinsic affinity to be separately engineered. Stability was optimized in the larger component, which has sufficient size to retain an independent tertiary structure. Intrinsic affinity was then

this approach for providing accurate results. Although qualitative correlation may be sufficient for certain experimental purposes, accurate representations of intracellular dynamics would normally be preferred. NanoLuc (Nluc) is an engineered luciferase derived from a deep sea luminous shrimp.9 The enzyme is small (19 kDa), stable, and produces bright and sustained luminescence. Taking advantage of these attributes, we sought to create a binary reporter system having minimal steric burden on the fusion partners but still providing high detection sensitivity. Our objective was to allow accurate quantitation of protein interactions under physiological conditions relevant to the cellular environment. Therefore, in addition to having a minimal physical presence, our reporter system was designed to have minimal influence on the affinity and association kinetics of the interacting target proteins. Furthermore, the reporter was 401

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology

Figure 2. Influence of NanoBiT on the SME1/BLIP interaction. (a) Inhibition profiles of the SME1/BLIP WT interaction with differing combinations of fused NanoBiT. Each profile was normalized by the β-lactamase (SME1) activity measured in the absence of BLIP. (b) Inhibition constants (Ki) determined by β-lactamase activity with different combinations of fused NanoBiT. Corresponding dissociation constants (KD) determined by luciferase activity where both SME1 and BLIP were fused to NanoBiT. (c) Relationship between NanoBiT luminescence and inhibition of SME1 (both normalized). (d) Increase in luminescence with time after the addition of BLIP-114 at various concentrations (0.5−30 nM). (e) Observed rate constants (kobs) for each concentration of BLIP-114. (f) Association (kon) and dissociation (koff) rate constants of BLIP binding to SME1 calculated from NanoBiT measurements. KD values calculated from the ratio of koff/kon. For all panels: n = 3, variability displayed as SD.

separately adjusted with the smaller component, which is too short to support a tertiary structure on its own.

Our results reveal that simply splitting a reporter protein may not necessarily provide optimal fragments for quantifying protein 402

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology

following inhibition of protein synthesis was monitored (Figure s5a,b). No significant effect on protein turnover was evident with either 11S or parental 156. A slight increase in degradation may be associated with attaching the unstructured NP (or a modified variant; Figure s5c−e). The apparent KD of 11S with NP was 900 nM (Figure 1f; Table s2), indicating that the affinity may be too high for use with weakly interacting proteins. Ideally, the intrinsic affinity of the complementation reporter should be substantially lower than the affinity of the interacting target proteins (with affinities in the high nM to μM range). To reduce the intrinsic affinity of the reporter, a semirandom library of 350 variants of NP comprising generally conservative amino acid substitutions (or deletions at each terminus) was synthesized and screened for affinity and luminescence with 11S. This uncovered complementing peptides with affinities that spannned 5 orders of magnitude, yet for many of the peptides the specific activity (i.e., Vmax) across the variants was relatively unaffected (Figure 1f; Table s2). The lowest affinity was achieved with 114, an 11-amino acid peptide (1.3 kDa) exhibiting an apparent KD of 190 μM. The luminescence spectrum of 11S with 114 is essentially unchanged from that of Nluc (peak emission ∼460 nm; Figure s6). The weak association between 11S and 114 indicates that they should not significantly alter the binding affinity of target proteins having KD values up to ∼10 μM. Further investigation revealed that peptides even shorter than 11 amino acids could deliver productive complementation with 11S, although with some loss of light intensity (Table s3; Figure s7). In addition, we found that a much shorter linker (Table s4; Figure s8) or no linker at all (Table s5; Figure s9) may be sufficient for the peptide. Having a shorter complementary peptide or linker may be beneficial in circumstances where a minimal fusion tag is preferred. Accuracy of Interaction Measurements. We used the interaction between purified SME-1 β-lactamase (SME1) and βlactamase inhibitory protein (BLIP)13 as a model to assess the accuracy of 11S combined with 114 (NanoBiT) for quantifying protein interactions. This model allowed binding interactions to be measured either through inhibition of β-lactamase or by NanoBiT luminescence. Moreover, using reported affinity mutants of BLIP,14 we could evaluate a range of binding interactions without altering the overall molecular geometry. Specifically, we used BLIP WT (Ki = 2.4 nM), BLIP Y50A (Ki = 32 nM), and BLIP R160A (Ki = 322 nM) to provide a ∼100-fold range in binding affinity. Evaluation of different linkage orientations revealed optimal performance with 11S appended to the C-terminus of SME1 and 114 appended to the C-terminus of BLIP. This configuration apparently imposed minimal steric constraints, since appending differing permutations of this configuration had little effect on the inhibition profile of SME1 with the various BLIPs (Figure 2a,b; Figure s10a,b). The estimated Ki values corresponded to published measurements,14 with small differences between each other (2-fold or less) associated mostly with fusions of 114 to the BLIPs. From the configuration containing both subunits of NanoBiT, the measurements of luminescence corresponded closely with the inhibition of β-lactamase activity (Figure 2c; Figure s10c,d). Values for Ki (β-lactamase inhibition) and KD (NanoBiT bioluminescence) were essentially indistinguishable within the precision of the assays (Figure 2b), indicating that NanoBiT provided an accurate assessment of protein interactions in this model system.

interactions. Rather, compromised structural integrity resulting from protein fragmentation can reduce the activity of the reporter and as a result decrease detection sensitivity. Additionally, the intrinsic affinity of the fragments can limit the dynamic range of the reporter response and potentially interfere with the interaction characteristics of the target proteins. By adapting the structure of these fragments to serve as components of a binary reporter system, we have improved their analytical proficiency for revealing dynamic intracellular processes occurring under physiological conditions. In experiments designed to evaluate quantitative concordance of the luminescent signal with protein interactions, this complementation reporter based on Nluc delivered accurate representations of both affinity and association kinetics. As the components of this optimized reporter are structurally distinct from Nluc and for nomenclature simplicity, we refer to the reporter as NanoBiT (shorthand for NanoLuc Binary Technology).



RESULTS AND DISCUSSION Engineering a Complementation Reporter. To identify potential dissection points in Nluc, we created circularly permutated10 variants in which the N- and C-termini were relocated within the protein structure, and the original termini were tethered by a cleavable peptide linker (i.e., containing a recognition sequence for TEV protease). Ninety variants, representing permutation sites distributed throughout the Nluc sequence, were generated and screened in cell lysates for luciferase activity. Measurements were made both before and after proteolytic cleavage to identify fragments that require a proximity constraint for maintaining luciferase activity (Figure s1a). By this method, a dissection point was found between residues 156 and 157 (Nluc CP-157; Figure 1a; Figure s1b−g), leading to an 18 kDa polypeptide fragment (termed 156) and a much smaller 13-amino acid (1.5 kDa) peptide (termed native peptide, or NP). The binding affinity (KD = 6 μM) between 156 and NP is comparable to that between fragments of a commonly used split firefly luciferase (Fluc;11 Figure s2), suggesting their potential suitability as a complementation reporter. However, 156 expressed poorly in cells (Figure 1b) and was rapidly degraded compared to full-length Nluc (Figure 1c). This was likely a result of compromised structural stability associated with fragmentation, as evidenced by reduced thermal stability of the luminescent activity (Figure 1d) and formation of insoluble aggregates in E. coli (Figure s3a). To improve stability, 156 was randomly mutated and screened for luminescence in E. coli lysates containing NP. Higher luminescence should result from increased enzyme expression or increased specific activity, either of which were expected to correlate with improvements in structural stability. Two rounds of mutagenesis and screening (nearly 15,000 variants) uncovered 16 beneficial amino acid substitutions which were combined to produce the variant, 11S (Figures s3, s4a; Table s1). In mammalian cells, 11S exhibited greater protein expression (Figure 1b) and reduced protein turnover (Figure 1c) to levels comparable with Nluc. Luminescence expression, assayed in the presence of NP, was also increased by over 300-fold (Figure s4b), with specific activity increased 3-fold (Figure 1e) to 37% of Nluc. Thermal stability of the luminescence (i.e., Tm) increased from 48 to 54 °C (Figure 1d). While appending an unstable polypeptide could potentially promote degradation of a target protein, it could also result in suppression of degradation. To investigate this, 11S and 156 were fused to Fluc-PEST (FlucP),9,12 and the loss of FlucP activity 403

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology

99%), consistent with essentially complete dissociation of FRB/ FKBP. As expected, the interaction profile in cell lysates shifts strongly to the right due to dilution of interacting proteins into the culture medium. The lower baseline results from reduced spontaneous luminescence (i.e., autoluminescence) generated by furimazine in the lytic reagent. Intracellular Kinetics. As cellular processes are generally reactive to environmental conditions, it is advantageous for the reporter to also indicate the dynamic character of protein interactions. This is exemplified by the transent interaction of βarrestin-2 (ARRB2) with the β2-adrenergic receptor (ADRB2), in contrast to the more stable interaction it makes with the arginine vasopressin receptor 2 (AVPR2).15 This kinetic distinction was apparent by expressing ADRB2-11S/114ARRB2 or AVPR2-114/11S-ARRB2 in HEK293T cells treated with the appropriate agonist (isoproterenol and vasopressin, respectively; Figure 4a). Temporal differences were also observed by bioluminescence imaging of HeLa cells expressing the NanoBiT fusions (Figure s14a−d). More kinetically challenging is the interaction of protein kinase A catalytic (PRKACA) and type 2A regulatory (PRKAR2A) subunits in response to intracellular cAMP concentrations. Previous studies have shown that dynamic changes in the association of these subunits can occur rapidly in response to applying external stimuli to the cells.15 We explored the ability of NanoBiT to represent these rapid dynamics by expressing 114-PRKACA and 11S-PRKAR2A in HEK293T cells, along with a luminescence-based biosensor for intracellular cAMP.16 By utilizing a different substrate (Fluc luciferin), the biosensor can be independently assayed within cells. Modulation of the endogenous ADRB2 receptor through progressive addition of an agonist (isoproterenol) and antagonist (propranolol), followed by stimulation of adenylate cyclase with forskolin, caused rapid changes in cAMP concentrations as indicated by the biosensor. Corresponding inverse changes were evident in NanoBiT luminescence, showing precise temporal coupling of the reporters in response to the underlying cellular dynamics (Figure 4b). As molecular kinetics are fundamentally linked to temperature, measurements at the relevant physiological temperature are required to accurately represent interaction dynamics. The response profile of protein kinase A represented by NanoBiT showed that both association and dissociation rates were much faster at 37 °C than 21 °C (i.e., ambient temperature; Figure 5a,b). The analogous experiment using split Fluc as the complementation reporter (i.e., PRKACA-FlucC/FlucNPRKAR2A) showed a marked deterioration of the data quality at physiological temperature. This is consistent with a previous report describing the unstable nature of split Fluc.17 Because the 114 peptide is unlikely to have conformationally defined structure in the absence of 11S, the suitable performance of NanoBiT at higher temperature was likely due to the thermal stability of the 11S subunit. Pharmacological Modulation of Protein Interactions. Because complementation reporters are frequently used to investigate the influence of synthetic molecules on protein interactions, we evaluated the suitability of NanoBiT for this type of application. The effect of intrinsic affinity between the complementation subunits was investigated by expressing FRB11S in HEK293T cells, together with FKBP fused to peptides with differing affinities to 11S. Our results indicate that the apparent potency of rapamycin to induce protein dimerization was minimally affected by the intrinsic affinity (range >100,000

To assess the ability to quantify interaction kinetics, we also used NanoBiT to estimate the association (kon) and dissociation (koff) rate constants for SME1 with the BLIPs. In contrast to the assay for β-lactamase (colorimetric conversion of nitrocefin), the instantaneous production of luminescence allows dynamic processes to be continuously monitored. The increase in luminescence upon combining BLIP with SME1 conformed well to a pseudo-first-order reaction (Figure 2d; Figure s11a,b), and kobs values correlated linearly with BLIP concentrations (Figure 2e; Figure s10c,d). The data indicate that reduced affinity of the BLIP mutants was achieved largely by increased koff, with relatively little effect on kon. Values of KD calculated by the ratio of koff to kon corresponded well to the estimates of KD made under steady-state conditions (Figure 2f). The intrinsic kon and koff rate constants for NanoBiT were determined to be ∼500 M−1 s−1 and 0.2 s−1, respectively (Figure s12). The BLIP having the lowest affinity (R160A, measured KD ∼ 1 μM) also had the lowest kon and highest koff values. As the kon for NanoBiT was ∼10-fold lower than for R160A and the koff ∼ 10-fold higher, the intrinsic kinetics of NanoBiT should not affect the interaction kinetics of R160A with SME1. The effect of NanoBiT should be even less for the higher affinity BLIPs. NanoBiT may begin to influence the apparent kinetics of protein interactions having unusually slow association rates or fast dissociation rates, which may underly very weak binding affinities (e.g., KD > 10 μM). Protein Interactions in Cells. For NanoBiT to produce quantitative data in living cells, its luminescence must be linear with the intracellular concentration of the binary complex. Linearity of Nluc luminescence with reporter concentration has previously been established biochemically.9 In both live cells and cell lysates, luminescence was also proportional to the amount of Nluc DNA used for transfection (Figure 3), although deviating

Figure 3. Linearity of luminescence in cells and cell lysates. HEK293T cells transfected with NanoBiT (equal amounts of FRB-11S and FKBP114) or Nluc DNA were treated (±) with 30 nM rapamycin (R) and measured for luminescence before or after cell lysis. n = 3 (NanoBiT) or n = 6 (Nluc), variability displayed as SD. Curve fits for Nluc and NanoBiT were generated based on the expected first order and second order relationships, respectively.

slightly at high expression. To assess linearity for NanoBiT, the rapamycin-inducible FRB/FKBP model was used (i.e, FRB-11S/ FKBP-114; Figure s13a,b). At higher DNA concentrations, the intracellular luminescence from NanoBiT in the presence of rapamycin followed a linear trend similar to Nluc. The signal departed significantly at lower concentrations, suggesting some dissociation of FRB/FKBP even in the presence of rapamycin. These data conform to a second order model for the concentration-dependent interaction of two proteins. Luminescence was greatly reduced in the absence of rapamycin (i.e., > 404

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology

Figure 4. Monitoring temporal interactions in cells. (a) ARRB2 interaction with fast or slowly recycling GPCRs. The interactions of ADRB2-11S/114ARRB2 and AVPR2-114/11S-ARRB2 were followed continuously in HEK293T cells at 37 °C after treatment at t = 1 min with isoproterenol (10 μM) or Arg8-vasopressin (100 nM), respectively. Individual S/B values were scaled by their respective dynamic ranges by subtracting the minimum value of the data set, then dividing by the maximum value. (b) Reversible interaction of 114-PRKACA/11S-PRKR2A followed continuously in HEK293T cells at 28 °C. 10 μM isoproterenol (Iso), propranolol (Pro), and forskolin (Fsk) were added sequentially to modulate endogenous ADRB2 (Iso and Pro) or activate endogenous adenylate cyclase (Fsk). Changes in intracellular cAMP concentration were monitored independently using GloSensor cAMP. Data were scaled by respective dynamic ranges as in panel a. For both panels: n = 3, variability displayed as SD.

Figure 5. Influence of temperature on monitoring PKA. Comparison of NanoBiT and split Fluc for monitoring PKA interaction dynamics at 21 °C (a) and 37 °C (b). HEK293T cells transiently expressing PRKACA and PRKAR2A fused to NanoBiT (50 pg DNA/well) or split Fluc (5 ng DNA/well) were treated as described in Figure 4b. Data were normalized as described for Figure 4b. For both panels: n = 3, variability displayed as SD. Data for other DNA concentrations can be found in Figure s15.

pathway in cells expressing wild type BRAF.19 This phenomenon has been monitored in cells previously using kinase domains modified with the CAAX membrane targeting motif.20 We surmised that the small size and stability of NanoBiT may allow drug-induced dimerization to be detected in the full-length proteins. In cells expressing BRAF-11S/CRAF-114, an inhibitordependent increase in dimerization was observed, with the calculated EC50 for each inhibitor in agreement with previously described potencies20 (Figure 7a). Cells expressing the NanoBiT fusions exhibited increased phosphorylation of MEK and ERK in correlation with drug-induced dimerization21 (Figure s18a,b). NanoBiT was also capable of differentiating between the reversible inhibitor, GDC 0879, and the irreversible inhibitor, AZ 628. Removal of GDC 0879 from the cells caused the luminescence to decrease, indicating dissociation of BRAF/ CRAF. In contrast, removal of AZ 628 had no effect, indicating that it remained bound to its target (Figure 7b). These results indicate that the intrinsic affinity of the NanoBiT subunits is suitable for quantification of both the potency and kinetics of drug-induced protein interactions.

examined; Figure 6a; Figure s16a−i), and the values achieved are consistent with published reports.18 This should be expected since compound potency is typically associated with binding affinity to its target, rather than interaction affinity within the target complex. Thus, potency should be minimally influenced by the choice of complementation reporter so long as the binding site for the compound is unaltered. In contrast, the dynamic response induced by rapamycin is strongly dependent on the intrinsic affinity of the reporter (Figure 6b). This arises primarily from increased luminescence of the sample lacking rapamycin, presumably due to increased spontaneous interaction promoted by the interaction affinity of the reporter subunits. Thus, the highest assay sensitivity was achieved with the low affinity 114 peptide (S/B = 1,200; Z′-factor =0.98). A similar analysis was performed with split Fluc (Figure s17). More pharmaceutically relevant is the interaction of BRAF with CRAF, which paradoxically is induced by inhibitors designed for oncogenic mutants of BRAF. Although intended to promote tumor regression by suppressing the RAS-RAFMEK-ERK signaling pathway, these inhibitors can activate the 405

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology

Figure 6. Influence of NanoBiT on rapamycin-inducible FRB/FKBP interaction in cells. (a) Rapamycin potency (EC50) for inducing the interaction between FRB-11S and FKBP fused to various peptides. (b) Induction of protein interactions with 30 nM rapamycin in HEK293T cells coexpressing FRB-11S and FKBP-peptide. For both panels: n = 4, variability displayed as SD.

conformational stability and low intrinsic affinity. Increased stability designed into the large component (11S) provides improved expression and reporter performance at physiological temperature. Low intrinsic affinity, achieved by adapting the small peptide component (114), minimizes its influence on the interaction characteristics of the target proteins. The intrinsic affinity and association constants determined for NanoBiT were found to be outside of the ranges typical for protein interactions. Thus, the luminescent signal from NanoBiT should provide an accurate indication of interaction dynamics (e.g., for proteins having KD < 10 μM), although caution is prudent when working with unusually weak or transient interactions. This capability was verified using SME1 in combination with a set of interacting inhibitory proteins (BLIP) of differing affinities. Appending the reporter had negligible influence on binding affinity, although as with any fusion tag, the possibility of steric interference for certain protein pairs should be considered. Nonetheless, the luminescent signal from NanoBiT correlated precisely with the independent assessment of protein association through SME1 activity. In mammalian cells, the luminescence produced by NanoBiT is well behaved and reflective of the interaction status of the reporter. For the interaction of FKB/ FRBP, it showed second-order behavior consistent with a dynamic binary system, while at higher concentrations favoring full association of the dimeric complex, it conformed to the linear characteristics of NanoLuc. The luminescent response was rapid and reversible without evident lag, indicating that NanoBiT can react nearly instantaneously to changing intracellular conditions. Furthermore, dynamic processes can be monitored for more than an hour due to the stable signal duration of the luminescent reaction. Our analysis indicates that NanoBiT fulfills the general expectation of a complementation reporter in mammalian cells, with biomolecular parameters generally suited for accurate portrayal of intracellular processes. In addition to analyzing protein interactions within cells, we also confirmed the ability to characterize pharmacological outcomes. Specifically, NanoBiT can reveal drug potency for induced protein interactions and the binding kinetics underlying drug action. Moreover, with the ability to substantially alter both size and intrinsic affinity of the peptide component in this binary reporter, we believe this luminescent system will provide additional applications yet to be realized.

Figure 7. Measuring inhibitor-induced BRAF/CRAF dimerization. (a) BRAF-11S and CRAF-114 dimerization in HCT116 cells following 2 h incubation with different BRAF inhibitors. Calculated EC50 values shown in legend. (b) Washout experiment showing reversibility of BRAF-11S/CRAF-114 interaction. HEK293T cells were treated with 1 μM of the reversible inhibitor, GDC 0879, or the covalent inhibitor, AZ 628, for 2 h. For washout samples, inhibitor was removed by media exchange and luminescence was measured 1 h later. Data represent the remaining luminescent signal normalized to t = 0 of the curve fit. For both panels: n = 4, variability displayed as SD.

Summary. NanoBiT is a complementation reporter designed for quantitative investigation of protein interaction dynamics under relevant physiological conditions. Benefiting from the small size and bright luminescence of NanoLuc, it provides detection at low intracellular concentrations with minimal steric interference on appended target proteins. Notably, one of the NanoBiT componenents is only 11 amino acids long, comparable to other peptide tags routinely used in protein analysis. The other NanoBIT component is also relatively small in terms of protein tags (e.g., two-thirds the size GFP). NanoBiT is distinct from other complementation reporters because its components are structurally optimized for high 406

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology



METHODS

RLUmax =

DNA Constructs, Protein Purification, and Lysate Preparation. See the Supporting Information. Nluc Dissection and Screening. CP Nluc variants were made using splicing by overlap extension (SOE).22 Each variant was designed so that former termini were joined by a 33-amino acid polypeptide linker containing a TEV protease sequence (underlined): GSSGGGSSGGEPTTENLYFQSDNGSSGGGSSGG. Variants were screened by incubating bacterial, wheat germ, or mammalian lysates with or without ProTEV Plus in 1X ProTEV buffer (Promega) at 30 °C for 60 min. Samples were diluted in DMEM and then further diluted 1:2 in Nano-Glo Luciferase Assay Reagent (Promega) for 5 min at ambient temperature before measuring luminescence with a GloMax-Multi+ Detection System (Promega). Sequence Optimization. DNA encoding Nluc amino acids 1−156 was mutated by error-prone PCR (Diversify PCR Random Mutagenesis, Clontech), subcloned into pF4Ag,9 and used to transform KRX E. coli. Lysates of 3,696 variant clones were screened (Freedom robotic workstation (Tecan)) by incubation with HaloTag-NP (KRX lysates) for 10 min with shaking. Nano-Glo Luciferase Assay Reagent was added, and luminescence was measured on a GENiosPro reader (Tecan). In addition to fully random mutagenesis, we attempted to engineer structural stability to 156 by reducing conformational entropy.23 Each glycine residue was substituted to alanine, and beneficial (∼1.5-fold improved luminescence) changes (G15A, G35A, G51A, G67A, and G71A) were combined with mutations identified from the random library screen. The combined variant providing the highest luminescence was used as the template for a second library, where 11,088 variants were screened in similar fashion. Improved variants identified during primary screening were validated using a secondary screen, where clones were processed the same way but in triplicate. Verified hits were then sequenced. Synthetic peptide arrays (≥75% purity; N-terminus acetylated and the C-terminus amidated; New England Peptide) were designed to semirandomly sample combinatorial sequence space (via generally conservative amino acid substitutions and deletions at each terminus) across NP. Peptides were dissolved in water and concentration quantified by A280. All other synthetic peptides of interest were synthesized at >95% purity (Peptide 2.0). Relative binding affinity and maximal luminescence were determined by titrating the peptide into large NanoBiT variants as described below under peptide affinities. Intracellular and Thermal Stability. HeLa cells expressing 156, 11S, or Nluc were plated at 104 cells/well (96-well plate, Corning), and 24 h later medium was exchanged with growth medium containing 0.4 mM cycloheximide or DMSO. Cells were assayed over 6 h by addition of Nano-Glo Luciferase Assay Reagent (with 100 μM NP for 156 and 11S) and measured on a GloMax-Multi+ Detection System. Cycloheximidetreated samples were normalized to DMSO-treated samples for each time point, and data were fit (one phase decay) using GraphPad Prism 6. Thermal stability analysis was carried out by incubating 1.5 pmol 156 or 30 fmol 11S (in DMEM containing 0.1% Prionex) for 30 min at different temperatures. Following rapid cooling to 4 °C, 300 pmol of synthetic NP was added, and samples were incubated at 22 °C for 5 min. Nano-Glo Luciferase Assay Reagent was added, and luminescence was measured using an Infinite F500 reader (Tecan). Activity was normalized to control samples incubated at 4 °C. Tm represents the temperature at which 50% of protein is inactive. Specific Activity. Purified Nluc, 156, and 11S were diluted to 40 pM in assay buffer (PBS pH 7, 0.01% Prionex, 0.005% Tergitol, 1 mM DTT). Synthetic NP was titrated (64 nM−140 μM for 156; 9 nM−20 μM for 11S; buffer in place of peptide for Nluc) and samples assayed with variable furimazine (Fz) concentrations (78 nM−10 μM for Nluc; 625 nM−20 μM for 156/11S) in Nano-Glo Buffer. For each concentration, maximal luminescence (RLUmax) was calculated using the equation,

RLU =

Vmax × [Fz] KM + [Fz].

Peptide Affinities. Purified 11S (40 pM) or 156 (400 pM) were mixed with various concentrations of synthetic peptide (Peptide 2.0) in assay buffer and incubated at ambient temperature for 30 min. NanoGlo Luciferase Assay Reagent was added and luminescence measured on a GloMax-Multi+ Detection System. Average RLUs over 10 min were fit (one site specific binding) using GraphPad Prism 6, and the fits were subsequently used to determine KD values. β-Lactamase Assays. Purified BLIP variants were incubated with SME1 for 2 h at ambient temperature. Nitrocefin (EMD Millipore) was added to the equilibrated complex, and substrate hydrolysis was monitored (A280) every 25 s for 25 min on an Infinite M1000 reader (Tecan). Using these measurements, inhibition constants were calculated as previously described.13 Luminescence-based Binding Assays. KD values were determined by adding Fz (10 μM) to the equilibrated complex, measuring luminescence on a GloMax-Multi+ Detection System every 1 min for 30 min, and fitting the data to the equation,

RLU =

RLUmax × [BLIP] KD + [BLIP].

For BLIP binding rate constants, BLIP-114, SME1-11S, and Fz were combined, and luminescence was monitored for 15 min (BLIP WT114) using a GloMax-Multi+ Detection Systemor every 1 s for 100 s (BLIP Y50A-114 and BLIP R160A-114) using the Infinite M1000 reader (Tecan). Observed rate constants (kobs) were determined by fitting (one phase association) with GraphPad Prism 6. The observed rate constant was then plotted versus BLIP-114 concentration, and the linear fit was determined. The association and dissociation rate constants were determined by fitting kobs data to the equation,

kobs = kon × [BLIP] + koff . Intracellular Luminescence. For each model, the eight possible orientations of the reporter subunits fused onto the interacting proteins were screened for maximal S/B. Using the optimal orientations, mammalian cells transiently expressing the fusion proteins (1:1 ratio of interacting pairs) were plated at 2 × 104 cells/well (96-well plate, Corning) and incubated ± drug at 37 °C. For FRB/FKBP, HEK293T cells were incubated for 2 h with rapamycin (EMD Millipore) in OptiMEMI. For BRAF/CRAF, HEK293T or HCT116 cells were serum starved for 4 h in Opti-MEMI before incubation for 1 h with GDC 0879 (R&D Systems), AZ 628 (R&D Systems), Sorafenib (Cell Signaling Technology), or PLX4720 (Cayman Chemical Company). Nano-Glo Live Cell Substrate (Promega) was added to a 1× concentration, and luminescence was measured using a GloMax-Multi+ Detection System. For kinetic measurements, HEK293T cells were plated at 1 × 104 cells/well and transfected the following day with ADRB2, AVPR2, ARRB2, PRKAR2A, or PRKACA DNA. Twenty-four hours posttransfection, growth medium was exchanged with Opti-MEMI, and cells were incubated for 3.5 h at 37 °C. Nano-Glo Live Cell Substrate was added, and luminescence was measured using a Varioskan Flash Multimode Reader (Thermo Fisher Scientific) at either 21, 28, or 37 °C. Following thermal equilibration, compound was added by injection to final concentrations of 10 μM (isoproterenol, propranolol, forskolin) or 100 nM (Arg8-vasopressin) in Opti-MEMI and measurements continued. Split Fluc assays were performed in the same manner except using 2 mM Luciferin-EF (Promega). The GloSensor cAMP Assay (Promega) was used to monitor the cAMP response following the manufacturer’s recommended protocol at variable temperatures and with the addition of 2% (v/v) GloSensor cAMP Reagent 30 min prior to beginning luminescence measurements. Curve fitting luminescence linearity in cells and lysates was modeled using the expected first order relationship between Nluc expression and RLUs (a[DNA]+b) and the expected second order relationship for NanoBiT (a[DNA]2/(k+[DNA])+b). Additional methods can be found in the Supporting Information.

RLUmax × [Peptide] KD + [Peptide].

Vmax was then calculated using the equation, 407

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408

Articles

ACS Chemical Biology



(11) Paulmurugan, R., and Gambhir, S. S. (2007) Combinatorial library screening for developing an improved split-firefly luciferase fragment-assisted complementation system for studying protein-protein interactions. Anal. Chem. 79, 2346−2353. (12) Swanson, B., Fan, F., and Wood, K. V. (2007) Enhanced response dynamics for transcription analysis using new pGL4 luciferase reporter vectors. Cell Notes 17, 3−5. (13) Zhang, Z., and Palzkill, T. (2003) Determinants of binding affinity and specificity for the interaction of TEM-1 and SME-1 beta-lactamase with beta-lactamase inhibitory protein. J. Biol. Chem. 278, 45706−45712. (14) Petrosino, J., Rudgers, G., Gilbert, H., and Palzkill, T. (1999) Contributions of aspartate 49 and phenylalanine 142 residues of a tight binding inhibitory protein of beta-lactamases. J. Biol. Chem. 274, 2394− 2400. (15) Oakley, R. H., Laporte, S. A., Holt, J. A., Barak, L. S., and Caron, M. G. (1999) Association of beta-arrestin with G protein-coupled receptors during clathrin-mediated endocytosis dictates the profile of receptor resensitization. J. Biol. Chem. 274, 32248−32257. (16) Binkowski, B. F., Butler, B. L., Stecha, P. F., Eggers, C. T., Otto, P., Zimmerman, K., Vidugiris, G., Wood, M. G., Encell, L. P., Fan, F., and Wood, K. V. (2011) A luminescent biosensor with increased dynamic range for intracellular cAMP. ACS Chem. Biol. 6, 1193−1197. (17) Ohmuro-Matsuyama, Y., Chung, C. I., and Ueda, H. (2013) Demonstration of protein-fragment complementation assay using purified firefly luciferase fragments. BMC Biotechnol. 13, 31. (18) Banaszynski, L. A., Liu, C. W., and Wandless, T. J. (2005) Characterization of the FKBP-rapamycin-FRB ternary complex. J. Am. Chem. Soc. 127, 4715−4721. (19) Holderfield, M., Nagel, T. E., and Stuart, D. D. (2014) Mechanism and consequences of RAF kinase activation by small-molecule inhibitors. Br. J. Cancer 111, 640−645. (20) Lavoie, H., Thevakumaran, N., Gavory, G., Li, J. J., Padeganeh, A., Guiral, S., Duchaine, J., Mao, D. Y., Bouvier, M., Sicheri, F., and Therrien, M. (2013) Inhibitors that stabilize a closed RAF kinase domain conformation induce dimerization. Nat. Chem. Biol. 9, 428−436. (21) Hatzivassiliou, G., Song, K., Yen, I., Brandhuber, B. J., Anderson, D. J., Alvarado, R., Ludlam, M. J., Stokoe, D., Gloor, S. L., Vigers, G., Morales, T., Aliagas, I., Liu, B., Sideris, S., Hoeflich, K. P., Jaiswal, B. S., Seshagiri, S., Koeppen, H., Belvin, M., Friedman, L. S., and Malek, S. (2010) RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth. Nature 464, 431−435. (22) Horton, R. M., Hunt, H. D., Ho, S. N., Pullen, J. K., and Pease, L. R. (1989) Engineering hybrid genes without the use of restriction enzymes: gene splicing by overlap extension. Gene 77, 61−68. (23) Scott, K. A., Alonso, D. O., Sato, S., Fersht, A. R., and Daggett, V. (2007) Conformational entropy of alanine versus glycine in protein denatured states. Proc. Natl. Acad. Sci. U. S. A. 104, 2661−2666.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acschembio.5b00753. Figures s1−s18, Tables s1−s5, and supporting methods (PDF)



AUTHOR INFORMATION

Corresponding Author

*Tel.: 608-274-1181. E-mail: [email protected]. Present Address

§ Department of Pharmaceutics and Pharmaceutical Chemistry, University of Utah, Salt Lake City, Utah 84112, United States

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank G. Vidugiris for help with screening sequence libraries and G. Colwell at Gene Dynamics, LLC, for assistance with vector constructions.



REFERENCES

(1) Braun, P. (2012) Interactome mapping for analysis of complex phenotypes: insights from benchmarking binary interaction assays. Proteomics 12, 1499−1518. (2) Michnick, S. W., Ear, P. H., Manderson, E. N., Remy, I., and Stefan, E. (2007) Universal strategies in research and drug discovery based on protein-fragment complementation assays. Nat. Rev. Drug Discovery 6, 569−582. (3) Magliery, T. J., Wilson, C. G., Pan, W., Mishler, D., Ghosh, I., Hamilton, A. D., and Regan, L. (2005) Detecting protein-protein interactions with a green fluorescent protein fragment reassembly trap: scope and mechanism. J. Am. Chem. Soc. 127, 146−157. (4) Rossi, F., Charlton, C. A., and Blau, H. M. (1997) Monitoring protein-protein interactions in intact eukaryotic cells by betagalactosidase complementation. Proc. Natl. Acad. Sci. U. S. A. 94, 8405−8410. (5) Stefan, E., Aquin, S., Berger, N., Landry, C. R., Nyfeler, B., Bouvier, M., and Michnick, S. W. (2007) Quantification of dynamic protein complexes using Renilla luciferase fragment complementation applied to protein kinase A activities in vivo. Proc. Natl. Acad. Sci. U. S. A. 104, 16916−16921. (6) Misawa, N., Kafi, A. K., Hattori, M., Miura, K., Masuda, K., and Ozawa, T. (2010) Rapid and high-sensitivity cell-based assays of protein-protein interactions using split click beetle luciferase complementation: an approach to the study of G-protein-coupled receptors. Anal. Chem. 82, 2552−2560. (7) Remy, I., and Michnick, S. W. (2006) A highly sensitive proteinprotein interaction assay based on Gaussia luciferase. Nat. Methods 3, 977−979. (8) Takakura, H., Hattori, M., Takeuchi, M., and Ozawa, T. (2012) Visualization and quantitative analysis of G protein-coupled receptorbeta-arrestin interaction in single cells and specific organs of living mice using split luciferase complementation. ACS Chem. Biol. 7, 901−910. (9) Hall, M. P., Unch, J., Binkowski, B. F., Valley, M. P., Butler, B. L., Wood, M. G., Otto, P., Zimmerman, K., Vidugiris, G., Machleidt, T., Robers, M. B., Benink, H. A., Eggers, C. T., Slater, M. R., Meisenheimer, P. L., Klaubert, D. H., Fan, F., Encell, L. P., and Wood, K. V. (2012) Engineered luciferase reporter from a deep sea shrimp utilizing a novel imidazopyrazinone substrate. ACS Chem. Biol. 7, 1848−1857. (10) Yu, Y., and Lutz, S. (2011) Circular permutation: a different way to engineer enzyme structure and function. Trends Biotechnol. 29, 18− 25. 408

DOI: 10.1021/acschembio.5b00753 ACS Chem. Biol. 2016, 11, 400−408