JOURNAL OF BIOMOLECULAR SCREENING
`Volume 6 Number 6 2001
`© The Society for Biomolecular Screening
`
`High Density Miniaturized Thermal Shift Assays as a
`General Strategy for Drug Discovery
`
`MICHAEL W PANTOLIANOt EUGENE C PETRELLA JOSEPH D KWASNOSKI
`VICTOR S LOBANOV JAMES MYSLIK EDWARD GRAF TED CARVER ERIC ASEL
`BARRY A SPRINGER PAMELA LANE and FR SALEMME
`
`ABSTRACT
`
`More general and universally applicable drug discovery assay technologies
`are needed in order to keep pace with
`chemistry and genomicsbased target generation Ligandinduced conforma
`the recent advances in combinatorial
`tional stabilization of proteins is a well understood phenomenon in which substrates inhibitors cofactors and even
`other proteins provide enhanced stability to proteins on binding This phenomenon is based on the energetic cou
`pling of the ligandbinding and protein melting reactions In an attempt
`to harness these biophysical properties for
`drug discovery fully automated instrumentation was designed and implemented to perform miniaturized fluores
`cence based thermal shift assays in a microplate format for the high throughput screening of compound libraries
`Validation of this process and instrumentation was achieved by investigating ligand binding to more than 100 pro
`tein targets The general applicability of the thermal shift screening strategy was found to be an important advan
`tage because it circumvents the need to design and retool new assays with each new therapeutic target Moreover
`the miniaturized thermal shift assay methodology does not require any prior knowledge of a therapeutic targets
`ideally suited for the quantitative high throughput drug screening and evaluation of targets de
`function making it
`rived from genomics
`
`INTRODUCTION
`
`CHALLENGE
`
`CURRENTLY
`
`facing conventional high
`
`AMAJOR
`throughput screening drug discovery methodologies is the
`for the ex
`to rapidly judge relative binding affinities
`ability
`panding numbers of new compounds derived from combinato
`libraries Compounding this task is the recent
`chemical
`flood of new therapeutic targets that have become available
`efforts in both human and
`in genomicsbased
`from advances
`microbial organisms Moreover many of the targets derived
`from genomics have unknown functions12 andor unknown in
`vivo ligands ie orphan receptors34 making conventional as
`say development and drug screening problematic Therefore it
`has become increasingly apparent that more general and uni
`versally applicable assays are needed in order to scale with the
`exponential growth of diverse compounds and targets Some
`examples of general assay strategies have begun to appear
`in
`
`the literature5 many of which employ electron spray mass spec
`trometry as the ligandbinding detection system67
`Thermal shift assays have many biophysical attributes that
`satisfy the requirements of a general cross target drug discov
`stabilization of pro
`conformational
`ery assay Ligandinduced
`teins is a well understood phenomenon in which substrates in
`hibitors cofactors metal
`analogs of natural
`ions synthetic
`ligands and even other proteins provide enhanced
`stability to
`proteins on binding811 This phenomenon is based on the en
`ergetic coupling of the ligandbinding and receptor melting re
`actions as depicted in Figure 1 This energetic linkage results
`in the midpoint for thermally in
`in liganddependent
`changes
`for the ligandreceptor complex relative
`duced melting curves
`to the uncomplexed receptor ATm that are directly propor
`to the ligand binding affinity Ka Traditionally thermal
`shift assays have been conducted
`using differential scanning
`calorimeters DSCs that monitor the change
`in heat capacity
`
`tional
`
`rial
`
`3 Dimensional Pharmaceuticals Inc Exton PA
`Innovative Research SBIR Grant R43 GM52786 Phases I
`Supported in
`part by Small Business
`General Medical Sciences at the National
`Institutes of Health
`Inc Cambridge MA
`
`IPresent address Millennium Pharmaceuticals
`
`II 19951998 from the National
`
`Institute of
`
`429
`
`from the SAGE Social Science Collections All Rights Res
`
`BioRad Exhibit 2008
`IPR201700055
`Page 1 of 12
`
`

`

`430
`
`PANTOLIANO ET AL
`
`IMO
`
`Folded
`
`Molten
`
`Globule
`
`unfolded
`
`with 1gartcl Bowl
`
`FIG 1 Energetics of ligandinduced conformational stabilization of proteins The energetics of thermal shift assays is based on hg
`and dependent receptor stabilization as monitored by ATT measurements
`complexes rel
`for the melting transitions of ligandreceptor
`stabilization is directly related to the ligand binding affinity AGbind because
`receptor This ligandreceptor
`ative to the uncomplexed
`of the thermodynamic linkage between the ligandreceptor binding and ligandreceptor melting free energy functions AGbind and AEI
`for most
`the molten globule is the first state encountered during protein phase transitions5052
`respectively
`is generally believed that
`proteins and the energetics depicted here reflect this phenomenology
`
`It
`
`undergo temperature induced melting
`transi
`as proteins
`tions812 In addition biophysical methods that employ tem
`instrumentation that measures tem
`perature regulated optical
`and
`in absorbance fluorescence
`
`perature dependent
`changes
`circular dichroism have also been historically used to perform
`thermal shift assays1315
`In order
`
`the nearly universal physicochemical
`to exploit
`properties of ligandinduced
`stabilization of proteins for appli
`cations in drug discovery we have miniaturized the thermal
`shift assay for use in a high density microplate format The val
`idation of this miniaturized thermal shift process along with
`the implementation of the instrumentation and software is the
`subject of this report
`
`EXPERIMENTAL PROCEDURES
`
`Proteins
`
`The human a estrogen receptor ESR was purchased from
`PanVera Madison WI Bacteriorhodopsin H halobium
`was purchased from Sigma Chemical Co St Louis MO
`Human a thrombin was purchased from Enzyme Research
`Labs South Bend IN Bovine liver dihydrofolate
`reductase
`DHFR was purchased from Sigma The extracellular do
`mains of the fibroblast growth factor receptor 1 DIIDIII
`FGFR11 were cloned from a human placenta library and ex
`
`pressed as a secreted protein from HEK293E cells details to
`be described elsewhere The PilD enzyme a leader peptidase
`from Pseudomonas aeruginosa
`and an Nmethyltransferase
`from Wim Hol of the University of Wash
`was a generous gift
`ington Seattle
`
`Fluorescent
`
`dyes
`
`dyes 1anilinonaphthalene8
`
`The environmentally sensitive
`sulfonic acid 18ANS 26ANS 2ptoluidiny1napthalene
`6sulfonic acid 26TNS and the 54dimethylaminopheny1
`24phenyloxazole family of dyes Dapoxyl° sulfonic acid
`from Molecular Probes Eugene OR Envi
`were purchased
`ronmentally sensitive dyes are fluorophores that have low quan
`turn yields low QY in solvents with high dielectric constants
`such as water but are highly fluorescent high QY in solvents
`with low dielectric constants such as ethanol and dioxane16
`Examples of these dyes include the napthylamine sulfonic acids
`such as 18ANS 26ANS and 26TNS and also the Dapoxyl
`family of dyes17 They can be used as extrinsic probes of un
`folded or melted proteins because the dyes partition themselves
`into the melted protein or molten globule states that have low
`dielectric properties resembling those of organic solvents The
`result is a large increase in fluorescence for the extrinsic dye
`as a function of protein thermal melting These dyes are ex
`region and have emission in the vis
`cited in the near ultraviolet
`
`ible region 460530 nm
`
`BioRad Exhibit 2008
`IPR201700055
`Page 2 of 12
`
`

`

`APPLICABILITY OF THERMAL SHIFT ASSAYS FOR DRUG DISCOVERY
`
`431
`
`Microplate thermal shift assay process
`and instrumentation
`
`The proteindye solutions were first dispensed into each well
`of a 384 well microplate polystyrene conical well microplates
`MJ Research Inc Waltham MA in equal 25 or 50 pt
`aliquots Then an equal volume of test compound solution typ
`in 510 DMSO were dispensed
`ically 100200 aull
`into
`each well to yield test ligands at 50100 uM in an assay volume
`of 50100 µL The resulting amount of protein utilized for an
`is 025050 µgwell 15 AM for
`average therapeutic
`an 300kDa protein or 100200 Agplate depending
`on the
`volume chosen Finally a small volume of mineral oil was added
`to each well
`to eliminate evaporation Reference
`control wells
`ligands containing the appropriate amount of DMSO
`no test
`were distributed on the 384 well plate usually in rows 12 and
`24 The plates were subsequently heated in the custom designed
`
`target
`
`microplate thermal shift assay instrument Fig 2 which auto
`matically monitored the complete thermal melting of the proteins
`device CCD camera detection of changes
`through closedcircuit
`in fluorescence
`for the extrinsic environmentally sensitive fluo
`rophore A Genesis 5000 liquid handling robot Tecan Group
`Ltd Mannedorf Switzerland controlled the fluidics dispensing
`of all
`reagents to assay microplates
`
`RESULTS
`
`Microplate thermal shift process and instrumentation
`
`A fully automated miniaturized
`thermal
`shift assay
`that
`works in a high density microplate format was designed and
`implemented so that 384 protein melting transitions are simul
`instruments were con
`taneously processed Pilot and beta
`
`A
`
`Data
`
`Analysis
`Computer
`
`Thermally
`Controlled
`Sample
`Plate
`
`B
`
`Fluorescent
`
`Reporter
`Low QY in High
`Dielectric Media
`ie 1120
`
`hVA
`
`hvF
`
`Heat
`
`T
`
`AG fold
`
`Native Protein
`
`Melted Protein
`Molten Globule
`
`Fluorescent
`
`Reporter
`High QY in Low
`Dielectric Medium
`
`FIG 2 A Microplate thermal shift
`instrumentation The hardware employed for this study utilized a noncooled CCD imaging camera
`that automatically captured the fluorescence changes resulting from the extrinsic dyes that correspond to the thermal melting transitions of
`proteins as they were simultaneously heated on a computer controlled heating block Microplate heating occurred for 1 to 3 min intervals
`in 010 to 50°C increments for any temperature range between 25° and 99°C The CCD images were collected at each designated timetem
`for analysis at the end of the temperature range Typically microplate images were
`perature point and then stored on a personal computer
`collected at 2040 temperature points Microplate well towell temperature uniformity was controlled to yield protein melting temperatures
`that did not exceed ±025°C for all 384 wells in the absence of ligands Epiillumination from long wave
`Tni with standard deviations
`light and a selectable band pass filter allowed the desired fluorescence emission wave
`lengths to reach the CCD camera B Fluorescence based detection of thermal phase transitions in proteins
`lamps provided
`the excitation
`length ultraviolet
`
`BioRad Exhibit 2008
`IPR201700055
`Page 3 of 12
`
`

`

`432
`
`in 15 h or 6000 as
`structed to perform 384 assays
`saysday and are described in the legend to Figure 2A
`The miniaturized thermal shift assay process begins with a
`purified therapeutic protein 75 pure sample at 100
`µgmL in a suitable buffer near neutral pH The protein solu
`tion is then made 100 AM in an environmentally
`sensitive flu
`orescent dye and dispensed into each well of a 384 well mi
`croplate in equal 25 or 50 AL aliquots The monitoring of
`was accomplished
`miniaturized thermal
`phase
`through the use of environmentally
`sensitive fluorescent dyes
`as depicted in Figure 2B A more detailed description of the
`microplate thermal shift process together with the attending
`hardware and software is found in Experimental Procedures
`and elsewhereI820
`
`transitions
`
`Validation studies
`
`and the instrumentation was at
`Validation of this process
`tempted by investigating ligand binding to more than 100 model
`and therapeutic target proteins The following subset was se
`lected from these model systems to illustrate the utility of minia
`turized thermal shift assays for diverse classes of proteins
`
`ceptors
`
`Nuclear hormone receptors ESR The nuclear hormone re
`are a superfamily of ligandactivated DNA binding
`transcription factors ESR is the prototype member of this class
`of receptors involved in the regulation of eucaryotic gene ex
`pression and affecting cellular proliferation and differentiation
`is composed of several do
`tissues2I This receptor
`in target
`for hormone binding22 DNA binding23 and
`mains important
`activation of transcription
`The thermal melting transition for ESR was observed to be at
`445°C in the absence of ligands but was increased to 498°C in
`the presence of the known antagonist tamoxifen Fig 3A Good
`agreement was found for the binding affinity measurement made
`by the miniaturized thermal shift assay Kd = 11 uM and that
`reported in the literature IC50 = 042 0424
`Often the pretransitional and posttransitional baselines yt and
`Yu respectively for the well containing the ligand do not match
`those for the control well This discrepancy has three compo
`nents 1 optical effects 2 liquid handling variation and 3
`quenching of signal as a result of the ligand The optical effects
`are derived from the limitations of using a CCD camera with epi
`illumination There are optical edge effects
`that result from the
`lens of the camera and there are optical perspective
`effects
`the fluorescence mea
`caused by the depth of the wells so that
`sured for wells near
`the middle of the plate is larger than that
`measured near the edges In addition the epiillumination yields
`variable excitation light across the plate which results in vari
`across the CCD camera chip can
`able emission Nonuniformity
`also contribute to the observed variability
`in the pre and post
`transitional baselines The liquidhandling variation introduces
`small differences in volumes of reagents that result in differences
`in observed absolute fluorescence
`intensity Also introduction of
`bubbles during reagent delivery will change the absolute fluo
`rescence intensity derived from different wells Some degree of
`quenching of the fluorescence
`signal derived from the environ
`mentally sensitive dyes occurs
`for a subset of test compounds
`However
`the observed
`variations in absolute fluorescence
`sig
`nal are not a problem for these miniaturized thermal shift assays
`
`PANTOLIANO ET AL
`
`because both the pre and posttransitional baselines yf and Yu
`are measured independently for each well Consequently the crit
`ical thermal melting parameters Tm and Aliu for each well were
`also determined independently within the intrawell boundaries
`of yf and Yu see Eqn 3 in the Appendix
`
`Growth factor and cytokine receptors FGFR1 The FGFR
`1 extracellular domains DIIDIII FGFR I have been shown
`for ligandbinding interactions2527 The
`to be important
`known ligand pentosan polysulfate a sulfated oligosaccharide
`was found to increase the Ti by 25°C Fig 3B which corre
`sponds to a Kd of 55 AM at 25°C when calculated using Equa
`tion 3 see Appendix This is in good agreement with a re
`ported Kd of 11 AM for this ligand measured using isothermal
`titration calorimetry27
`
`Integral membrane proteins The utility of the miniaturized
`thermal shift assay strategy as a drug discovery tool for the in
`tegral membrane protein class was demonstrated with the fol
`lowing examples
`
`The enzyme PilD is
`PILD FROM PSEUDOMONAS AERUGINOSA
`enzyme with two catalytic ac
`a putative fivetransmembrane
`a leader peptidase and an Nmethyltransferase
`tivities
`activity
`that are essential for the proper processing of the type IV pili
`of many gram negative
`proteins expressed on the surface
`The results for the miniaturized thermal shift
`pathogens2829
`assay of ligands binding to PilD are shown in Figure 3C The
`activity cofactor SadenosylLmethionine
`Nmethyltransferase
`was found to increase the Tm by 11°C which corresponds to
`a binding affinity of Kd = 900 LtM This is in good agreement
`with the Kd of 300 AM previously measured using a ra
`dioactive competition assay
`
`BACTERIORHODOPSIN
`The seventransmembrane protein bac
`teriorhodopsin is a light driven proton pump from bacterial pur
`ple membranes and is commonly used as a model
`for struc
`of G proteincoupled
`turefunction
`studies
`receptors
`GPCRs31 The form of bacteriorhodopsin with retinal cova
`lent binding to Lys 216 was reconstituted in nonionic deter
`gents and investigated for its compatibility with the miniatur
`ized thermal shift process The melting transition for
`this
`to occur at 733°C as
`receptorligand complex was observed
`shown in Figure 3D
`
`natorial
`
`Screening of combinatorial
`libraries directed at the serine
`protease human a thrombin A series of seven focused combi
`libraries directed against the catalytic site of the serine
`protease a human thrombin were first assayed using a spec
`trophotometric enzyme assay and then compared with results
`
`One of four different
`receptors FGFR1 through FGFR425 that bind
`fibroblast growth factors a family of 18 related polypeptides FGF I through
`FGF1826 with broad mitogenic and cell survival activities These recep
`tors have been demonstrated
`to be important for embryonic development
`angiogenesis and wound healing in adults
`The enzymatic assays employ spectrophotometric substrates
`such as
`succinylAlaAlaProArgpnitroanilide
`which releases the colored prod
`uct pnitroanilide on hydrolysis
`The enthalpy measurement
`in the miniaturized thermal shift assay is
`of a Vant Hoff enthalpy determination
`the equivalent
`
`BioRad Exhibit 2008
`IPR201700055
`Page 4 of 12
`
`

`

`APPLICABILITY OF THERMAL SHIFT ASSAYS FOR DRUG DISCOVERY
`
`433
`
`a Tm = 55 5 A5
`o Tm = 58O B6
`
`Native D11 DlII FGFR11
`T = 555 °C
`
`on
`
`Da
`
`90
`
`e ad°u3Lia12
`ail 1
`
`csit
`
`o z°ciFt4ba
`
`7
`
`100 uM Pentosan Polysulfate
`AT tx 25°C
`
`50
`
`80
`
`70
`
`80
`
`80
`
`Temperature
`
`1
`
`4T
`
`Tm = 733
`
`Native Bacteriorhodopsin
`T= 733 °C
`
`A
`
`5500
`
`5000
`
`4500
`
`4o00
`
`£3500
`
`3000
`
`2500
`
`2000
`
`o Tm = 445 All
`o Tm = 498A6
`
`Native Estrogen
`Receptor
`T= 445 °C
`
`+ LJrovivr
`
`arrffsTr
`
`+++++
`
`3dirjalb+++++
`
`+++
`+++
`
`++
`
`Ya1
`
`lti
`
`I
`
`r3
`
`50 µM Tamoxifen
`AT = 53 °C
`
`25 3 35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`70
`
`75
`
`Temperature
`
`3400
`
`3200
`
`3003
`
`2000
`
`au
`
`2400
`
`2000
`
`1800
`
`0 Tm = 497 Al
`0 Tm = 508 A2
`
`Native P110
`Tin = 497 °C
`
`11200
`
`11000
`
`113300
`
`1 10600
`
`10400
`
`10200
`
`10000
`
`9800
`
`100 mM SAdenosylLMethionine
`Arm = 11 °C
`
`11000
`
`10000
`
`9000
`
`18000
`g
`
`7003
`
`60001
`
`5000
`
`40001
`
`35
`
`ill
`
`45
`
`40
`
`45
`
`60
`
`65
`
`70
`
`40
`
`50
`
`60
`
`70
`
`BO
`
`90
`
`100
`
`Temperature
`
`Temperature
`
`FIG 3 Miniaturized thermal shift assay results for diverse classes of enzymes and receptors A Ligand binding to the human a es
`trogen receptor The estrogen receptor was diluted into 50 mM Tris pH 80 80 mM KC1 16 glycerol 10 mM EDTA 10 sodium
`2 mM dithiothreitol
`and 10 mM MgC12 buffer which also contained
`100 104 of the fluorophore Dapoxyl sulfonic acid
`vanandate
`using the instrumentation described Figure 2 The known antagonist
`Fluorescence
`intensity was measured as a function of temperature
`tamoxifen was present at 50 pM and the final assay volume was 100 AL The solid lines through each of the data sets were fit using
`Equation 1 The binding affinity IQ at 25°C was calculated using Equation 3 and was found to be 11 iuM for tamoxifen An IC50 of
`042 iuM for tamoxifen was previously measured by fluorescence polarization24 B Ligand binding to fibroblast growth factor recep
`tor1 Miniaturized thermal shift assay data were obtained for DIIDIII FGFR1 in the presence and absence of ligands The mid
`point for the unfolding transition in the absence of ligands was observed to be 555°C while a known ligand pentosan polysulfate was
`the same conditions The calculated Kd at 25°C was found to be 55 cM Eqn 3
`observed to give a 25°C increase in Tn under
`which compares favorably with the Kd of 11 uM measured through isothermal titration calorimetry The buffer conditions for the as
`says were 50 mM phosphate buffered saline with 40 mM sodium citrate pH 72 10 mM MgC12 2 mM EDTA and 2 glycerol C
`enzyme PilD an example of an integral membrane protein Miniaturized thermal shift assays were
`Ligand binding to the processing
`performed for PilD in the presence and absence of ligands The melting transition in the absence of ligands 100 mM HEPES pH 751
`and 200 mM NaC1 in 123 mM noctyl6oglucopyranoside 05 CMCH20 was observed to have a Tm of 497°C The known
`100 mM is a cofactor
`for the methyl transferase activity of this enzyme and was observed to give
`ligand SadenosylLmethionine
`to a calculated Kd of 900 mM at 25°C Eqn 3 and compares
`a 11°C increase in Tm This corresponds
`favorably with the Kd of 500
`mM reported in the literature The assays were performed as described above with 2 Agwell PilD D Bacteriorhodopsin an example
`of a seventransmembrane protein Miniaturized thermal shift assays were performed for the retinal bound form of bacteriorhodopsin
`from H halobium The protein was solubilized in 5 mM Na0Ac pH 50 buffer containing 64 mM nnonylBDglucopyranoside
`CMCH20 The melting transition for this retinal bound covalent complex was observed to have a Tn of 733°C The assays were
`performed as described above
`
`BioRad Exhibit 2008
`IPR201700055
`Page 5 of 12
`
`

`

`434
`
`PANTOLIANO ET AL
`
`obtained using the miniaturized thermal shift process Fig 4
`The seven focused
`libraries consisted of seven 96 well plates
`synthesized in sequential order There were 213 hits out of the
`original 665 compounds with binding affinities that span 35
`orders of magnitude
`The results from the miniaturized thermal shift assay were
`found to correlate with the results obtained through the enzy
`to the slope =
`matic assay as demonstrated by the reasonable fit
`10 line for the loglog plot The deviation of the data from a
`fit can come from a number of sources the first of which
`
`perfect
`
`is the time interval 6 months between the enzyme and ther
`
`mal shift assays Time dependent
`for some labile li
`changes
`for some differences between the two meth
`gands could account
`ods Another source of some of the deviation may come from
`the range of ligandbinding enthalpies HL = 00 to 250
`kcalmol one might expect for an assortment of different
`li
`gands binding to a given target For examples of compounds
`that have binding affinities that are entropically driven see refs
`12 and 32 This potential assortment
`for ligandbinding en
`thalpies translates into a range for the calculated Ka at T for
`any given ligand when estimated using Equation 3 see Ap
`pendix For example the use of thrombin a ligand that yields
`a AT of 20°C results in a calculated Ka at 37°C of 85 X 105
`when AHL = 00 Eqn 3 while a ligand with AHL = 250
`
`kcalmol yields a calculated Ka at 37°C of 71 X 106 Thus
`in the case of thrombin a 10 fold range in the estimated Ka is
`possible for a given value of ATa which is consistent with the
`variation in Figure 4 Finally there could be a small
`observed
`subset of compounds that that interact with both the native and
`
`if
`
`the ligand
`
`unfolded forms of thrombin thereby resulting in smaller T
`determinations and smaller Ka than expected
`bound only to the native form of thrombin
`Data analysis for these screening experiments was simplified
`using a software interface that graphically displays the com
`puted A T determinations for all 384 wells of the microplate
`thermal shift assay as shown in Figure 5 The results for one
`of the library plates Plate 6 screened against
`thrombin is
`shown using an adjustable gradient color code for LITa mea
`sured for each wellthe more positive the
`TT for an indi
`vidual compound the more deeply redshaded the well White
`signifies little or no change in Tn compared with a control T0
`in which no compounds are present Thus hits can be visually
`identified and ranked by color and bar graph or alternatively
`they may be listed in tables that show the ATm values and es
`timated binding affinities Ka values for all 384 wells not
`shown More details concerning the data analysis software
`tools are located in the Appendix
`The software interface that graphically displays the computed
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`Log 1K1
`
`37°C Enzyme Assay
`
`10
`
`©37°CMiniaturizedThermalShift
`
`Ka
`
`Log
`
`ial
`
`libraries directed at
`
`assay
`
`FIG 4 Screening of focused combinatorial
`site of human a thrombin Seven focused combinator
`libraries directed at the catalytic
`site of human thrombin were first assayed by a conventional
`the catalytic
`enzymatic spectrophotometric
`and then compared with the results obtained using the miniaturized thermal shift process The seven focused libraries consisted of seven
`96 well plates 1 through 7 synthesized
`red diamonds = plate 1 blue boxes = plate 2 magenta boxes = plate 3
`in sequential order
`cyan triangles = plate 4 brown circles = plate 5 green diamonds = plate 6 black triangles = plate 7 The results were plotted in a
`loglog format for direct comparison of the extent
`to which the two different assay measurements are consistent The solid line of slope
`10 represents a perfect match for the two methods of analysis Thrombin was present at 10 AM in 50 mM Hepes pH 75 and 01 M
`NaCl and the Dapoxyl sulfonic acid was present at 100 AM Excitation of the fluorophore was effected using a long wavelength ul
`that allowed 500 to 530nm wavelength
`lamp and the emission was filtered through a bandpass filter
`to pass to the
`of 17 lt14
`CCD camera The compounds
`from the focused libraries were present at a final concentration
`
`traviolet
`
`light
`
`BioRad Exhibit 2008
`IPR201700055
`Page 6 of 12
`
`

`

`APPLICABILITY OF THERMAL SHIFT ASSAYS FOR DRUG DISCOVERY
`
`435
`
`1
`
`Hrrrrrnrr
`
`firtif14 44 10 17 18
`2 3 4 1 6 71$ V
`A lir 71171111711111111Tur 1701
`trrFrErniorrrrlrttatrrStrti
`c EFEEFFT1111111EEFFEIEFEll
`rrr711111rrriltrirrrrarrrrilrr
`F rrrrnrigarrrrnrrrruristrrr
`srrEarrinrrrrFrnorrnrrirn
`itrrrrrrrrrruriErrprrrrr
`Jrnrrnrrnrrnrrrrrrrnrrr
`K rrrrnrrsrurrrrrrrrmarrircr
`L rrnntrrrrwrrrrrrorrrrrplr
`1111111111111 ErrErarutrEnt
`htfT111111l
`N rrryrrirmrrrrrrnrrnrrtr
`o rrrurrnrrnrnnotrrrrrrrr
`P ErrrorrnrFirrrnarrrrrrar
`
`1000
`
`3503
`
`3X0
`
`I 2930
`
`1600
`
`1W3
`
`goo
`
`Raw Data
`
`Tra 629 P121
`Ta
`572 1613
`
`a
`
`gxPR
`
`114Pfq01
`
`26
`
`33
`
`36
`
`40
`
`46
`
`gt
`
`SS
`
`60
`
`65
`
`73
`
`Temparature
`
`library plates plate 6 screened against
`FIG 5 Data analysis software tools The ATm data analysis for one of the seven compound
`human thrombin Fig 4 is shown in various
`levels of detail The software used to analyze the miniaturized thermal shift assays per
`formed in 384 well plates displays the ATm values as an adjustable gradient color code shown at
`the right of the simulated 384 well
`the more negative the ATm White signifies
`the more positive the ATm Similarly the more blue the well
`plate The more red the well
`little or no change in Tm compared with control T0 The original 96 well plate of compounds was replicated four times on the 384
`the level of reproducibility of these measurements Al A13 Ii and 113 have the identical
`well assay plate in a quad format to reveal
`compound The control wells which contain only DMSO are designated as green which for this plate are wells H12 H24 P12 and
`P24 The fluorescence
`data for any well can be displayed for demonstration purposes by double clicking
`intensity versus temperature
`the right for well M23 which is compared with the data for a control P12 This compound
`on any well An example is shown at
`shows
`a ATm of 43°C and is one of the 21 hits on this plate with ATm 05°C A threedimensional bar chart
`representation of the A Tm data
`for this plate is also shown at the bottom The software was written in C+ + and runs on all Microsoft WindowsTmbased
`platforms
`
`Arm determinations for all 384 wells of the microplate thermal
`shift assay is described in the legend to Figure 5
`
`Multiligand binding events A demonstration of the effects
`of multiligand binding events on the miniaturized thermal shift
`assay appears in Table 1 Binding of the coenzyme NADP+
`and the folate mimetic methotrexate MTX to the bovine liver
`enzyme DHFR was analyzed individually as well as together
`ligands produced ATm values of 63°C
`Alone the individual
`for NADP± and 77°C for MTX which correspond to binding
`affinities of Kd = 24 AM for NADP+ and 58 nM for MTX
`at 25°C when calculated using Equation 3 The measurement
`for MTX is in agreement with those appearing in the literature
`see Table 1 Importantly when both ligands are present to
`there is a much larger increase in ATm of 181°C indi
`stabilization is to some extent
`the ligandinduced
`cating that
`is 41°C larger
`than that
`additive although the observed shift
`for the simple sum of the ATm values
`for the indi
`expected
`ligands This deviation from perfect additivity suggests
`
`gether
`
`vidual
`
`for these two ligands binding at ad
`some positive cooperativity
`jacent sites Support for this supposition comes from the struc
`for positive cooperativity of the
`tural and functional evidence
`binding of these two ligands at adjacent sites in DHFR enzymes
`from both eukaryotic and prokaryotic organisms3335 For ex
`ample a 100 fold increase in binding affinity of MTX at 10°C
`was reported when NADPH was bound to the bovine liver
`DHFR36 see Table 1
`
`DISCUSSION
`
`General drug discovery applications
`
`The successful measurements of binding affinity for assorted
`ligands binding to diverse classes of protein targets Figs 3 and
`4 demonstrate that
`the miniaturized thermal shift assay strat
`egy has wide cross target utility Of the 100 proteinligand
`complexes investigated to date only proteins with heme groups
`
`BioRad Exhibit 2008
`IPR201700055
`Page 7 of 12
`
`

`

`436
`
`Ligands
`
`None
`
`10 mM
`NADP
`NADPH 10 mM
`MTX 01 mM
`MTX
`NADP±
`
`NADPH
`
`MTX
`
`None
`
`NADPH
`MTX
`
`NADPH
`
`TABLE 1 EVALUATION
`
`BINDING USING MINIATURIZED
`OF MULTILIGAND
`BOVINE LIVER DHFR AS A TEST CASEa
`
`THERMAL
`
`SHIFT ASSAYS
`
`PANTOLIANO ET AL
`
`AT7 °C
`
`Kd nM
`
`Miniaturized Thermal Shift Data
`
`Tn = 518 ± 04
`63 ± 01
`
`97 ± 02
`
`77 ± 04
`
`181 ± 04
`= 140
`
`246 ± 06
`= 174
`
`Tn = 491
`
`82
`
`133
`
`2400
`
`810
`
`58
`
`Positive cooperativity
`Tm = +41°C >
`
`Positive cooperativity
`Tm = +72°C >
`
`DSC Data
`
`Kd nM in
`
`literature
`
`No data
`
`40 humanb
`33
`
`Positive
`
`cooperativity
`
`MTX binds 125
`fold tighter
`E colid
`
`Positive cooperativity
`MTX binds 100
`fold tighter
`Kd = 0033 nM`
`
`See ref 35
`
`See ref 35
`
`See ref 35
`
`See ref 35
`
`MTX
`
`290
`
`= 215
`
`Positive cooperativity
`= +75°C >
`
`The cooperativity of ligand binding to bovine
`
`liver DHFR was investigated
`using the miniaturized thermal shift assay Conditions employed were
`sum of individual ATn values for two ligands when present alone ie MTX
`
`the Kd for the human enzyme was reported to be 40 nM by Schweitzer
`
`50 mM HEPES pH 75 and 100 mM NaC1 in a volume of 5 t
`and NADPH under
`identical conditions
`bA Kd for NADPH binding to bovine liver DHFR could not be found however
`et al33
`
`eGilli et al36 These conditions were for bovine liver DHFR at pH 68 01 M potassium phosphate buffer at 10°C
`dCooperativity for NADPHr and MTX binding to Escherichia coil DHFR34
`Differential scanning calorimetry DSC conditions employed by Sasso et al33 were pH 68 01 M potassium phosphate buffer
`
`to yield low fluorescence as a result of quench
`were observed
`ing of the extrinsic dye fluorescence signal although this did
`not preclude using the assay at a higher concentration
`The binding affinity results obtained for the miniaturized
`thermal shift assay were found to be largely consistent with the
`fluorescence
`results obtained through enzymatic
`radioactive
`polarization isothermal
`titration calorimetry or other kinds of
`assays This was best demonstrated by the correlation between
`the continuous spectrometric enzyme assay and thermal shift
`data for the combinatorial
`libraries directed at the catalytic site
`of thrombin Fig 4
`One advantage of the miniaturized thermal shift strategy is the
`fundamental biophysical basis of the assay which circumvents
`the need to design and retool new assays for each new protein
`is that this strategy does not re
`target Another related outcome
`targets func
`quire any prior knowledge of a specific therapeutic
`tion in order to screen libraries and optimize drug leads In this
`regard the results for bacteriorhodopsin and the estrogen recep
`that the miniaturized thermal shift assay strategy will
`tor suggest
`be useful for drug discovery efforts with GECRs and nuclear hor
`
`mone receptors especially those orphan receptors for which drug
`screening presents a problem because of the lack of known lig
`ands These properties make a miniaturized thermal shift assay
`and
`ideally suited for high throughput drug screening
`strategy
`evaluation of targets derived from genomics
`
`Multiligand binding applications
`
`The data in Table 1
`illustrate the utility of the miniaturized
`thermal shift assay for assessing multiligand binding to protein
`targets The data show that this process
`can not only identify
`ligands binding to substrate binding sites of enzymes but also
`identify and judge the cooperativity of ligands binding to co
`factor allosteric andor regulatory sites on therapeutic targets
`Exosites on enzymes are often useful targets of therapeutic in
`tervention as demonstrated by the thrombin cofactor heparin
`an effective anticoagulant3738 In this regard the miniaturized
`thermal shift assay could be used in conjunction with protein
`information nuclear magnetic resonance
`flTMR1 or
`structural
`xray diffraction and be used to identify ligands binding at ad
`
`BioRad Exhibit 2008
`IPR201700055
`P