© 2019 Artel1

© 2013 ARTEL

SLAS2019 Tutorial: Coupling Assay Design And

Process Optimization Toward Minimizing Variability

Nathaniel Hentz, PhD

05 February; 12:30-1:15P

Acknowledgment: Becky Kitchener, PhD

© 2019 Artel2

© 2013 ARTEL

Tutorial Agenda

▪ Assay variability

▪ Liquid handler optimization: current practices

▪ Process optimization – What? Why? How?

▪ Case study: model assay & findings

▪ Conclusions

© 2019 Artel3

© 2013 ARTEL

OVERVIEW: ASSAY VARIABILITY

© 2019 Artel4

© 2013 ARTEL

Assay

Detector

ALH

Settings

Sample

Environment

Hardware /

Labware

Plate type, tip type,

tubing, automation

Off-set volume, single/multi

dispense, aspirate/dispense

rate, aspirate/dispense

height, liquid class, pre / post

air gaps, etc.

PMT, wavelength, x-y-z

position

Mixing, incubation

time, centrifugation, #

liquid transfers, wash

steps, dilution steps,

serial dilutions

Viscosity, density,

surface tension,

volatility

Temperature,

humidity, light,

vibration

Parameters that Effect Assay Variability

© 2019 Artel5

© 2013 ARTEL

Assay Validation in a High-Throughput World

Reagent

Stability

Signal

Variability

Plate

Uniformity

DMSO

Compatibility

Reaction

Stability

Assay

© 2019 Artel6

© 2013 ARTEL

So, What Does A “Good” Assay Look Like?

Let’s define a couple of parameters:

▪ First, let’s consider the assay variability

▪ Second, let’s consider the assay window

−

+−=

minmax

minmax 331

Z

Z-Factor Value Assay Quality

1 Ideal assay

1 > Z ≥ 0.5 Excellent assay

0.5 > Z > 0 Marginal assay

0 Yes/no assay

<0 Assay not useful

© 2019 Artel7

© 2013 ARTEL

The Result

Z’ = 0.74

0 1 2 2 4 3 6 4 8 6 0 7 2 8 4 9 6

0

2 0 0 0

4 0 0 0

6 0 0 0

8 0 0 0

1 0 0 0 0

W e ll ID

Sig

na

l (R

FU

)

Z’ = 0.94

0 1 2 2 4 3 6 4 8 6 0 7 2 8 4 9 60

2 0 0 0

4 0 0 0

6 0 0 0

8 0 0 0

1 0 0 0 0

W e ll ID

Sig

na

l (R

FU

)

© 2019 Artel8

© 2013 ARTEL

-2 -1 0 1 20

1000

2000

3000

4000

Log Inhibitor (nM)

Sig

na

l (

RF

U)

Opt 1

Opt 2

Opt 3

The Result

Useful assay descriptors:

Hill slope

Assay span

Upper asymptote

Lower asymptote

Variability

© 2019 Artel9

© 2013 ARTEL

OVERVIEW:

LIQUID HANDLER OPTIMIZATION

© 2019 Artel10

© 2013 ARTEL

Liquid Dispense Technologies

Dispense Technology Attributes

Air displacement

• Problematic for volatile liquids

• Possible cross contamination

• Wide range of volumes

Positive displacement

• Useful for volatile solvents

• Cross contamination possible

• Wide range of volumes

Droplet (acoustic)• Non-contact

• Useful for small volumes (pL – nL)

Droplet (solenoid/inkjet)• Useful for small volumes (nL)

• Sensitive to fluid types

Capillary (pintool)

• Useful for small volumes

• Direct contact with sample

(contamination)

• Sensitive to fluid type

Peristaltic• Useful for bulk dispense;

• More frequent calibrations needed

© 2019 Artel11

© 2013 ARTEL

Parameters that Effect Volume Transfer

▪ Hardware/labware – plate format, tip types, tubing type

▪ ALH settings – target (or off-set) volume, single/multi dispense,

aspirate/dispense rate, aspirate/dispense height, liquid class, pre and post air

gaps, accuracy/precision of volume transfer, transfer speed/time delays, on-

board mixing

▪ Assay – reagent mixing, incubation, centrifugation, number of liquid transfers,

wash steps, dilution steps, serial dilutions

▪ Sample – viscosity, density, surface tension, temperature, volatility

▪ Environment – temperature, humidity, light, vibration

© 2019 Artel12

© 2013 ARTEL

Typical Liquid Handler Optimization

▪ Usually performed as a stand-alone activity

▪ Precision is always checked, but accuracy is not as easy

▪ Several methods for volume verification

▪ In-house (e.g., fluorescence, gravimetric, absorbance, etc.)

▪ Commercial (e.g., dual-dye spectrophotometry)

▪ Volume verification is typically performed with ideal solutions

▪ Liquid handler is certified, calibrated, or repaired (if necessary)

▪ Then….someone programs ALH for assay use

▪ Default method

▪ Specific to basic assay requirements

© 2019 Artel13

© 2013 ARTEL

Artel MVS: A Useful Tool

▪ Employs a dual-dye, dual-wavelength,

ratiometric absorbance-based

measurement method for calculating the

dispense volume.

▪ How it works: dyes of known concentration

are dispensed into well-characterized

microtiter plate. The plate is mixed on a

plate shaker to ensure solution

homogeneity. Absorbance readings are

taken at 520 nm and 730 nm.

=

730

520

r

bTS

A

A

a

aVV

© 2019 Artel14

© 2013 ARTEL

ADOPTING A PROCESS OPTIMIZATION APPROACH

© 2019 Artel15

© 2013 ARTEL

Instrumentation

▪ Liquid handlers

▪ Detectors

▪ Mixers

▪ Incubators

▪ Centrifuges

Consumables

▪ Reagents

▪ Tips

▪ Plate type

▪ Plate seals

Sources of Assay Variability

Biology

▪ Diffusion

▪ Binding equilibrium

▪ Steric hindrance

▪ Cell population

diversity

▪ Protein activity

Environment

▪ Temperature

▪ Light

▪ Humidity

▪ People

What can

we control or

optimize?

© 2019 Artel16

© 2013 ARTEL

Artel MVS is More Than a Calibration Tool

Trouble-

shooting

Serial

Dilutions

Mixing

Plate

Washing

Tip

Evaluation

Calibrations

Plate

Evaluation

© 2019 Artel17

© 2013 ARTEL

CASE STUDY

© 2019 Artel18

© 2013 ARTEL

Assessing Effects of Liquid Handling on the Assay

▪ Phase I: Using a well-characterized model assay, develop and optimize the

assay platform at bench scale.

▪ Phase II: Perform method transfer to ALH: examine effects of automated

liquid handling parameters on the same assay.

▪ Phase III: “Deconstruct” the assay: decide which parameters, when altered,

significantly vary assay outcome.

© 2019 Artel19

© 2013 ARTEL

Streptavidin: Biotin-Fl Assay Principle

Streptavidin (SA) is a tetravalent biotin-binding protein that is isolated from

Streptomyces avidinii and has a mass of 60.0 kDa. SA has a very high affinity

for biotin (Kd = 10-14 to -15 M).

1. Waner, MJ; Mascotti DP. Journal of Biochemical and Biophysical Methods 70(6), 2008, 873-877.

2. Ebner, A; Marek, M; Kaiser, K; Kada, G; Hahn, CD; Lackner, B; Gruber, HJ. Methods in

Molecular Biology, 418, 2008, 73-88.

Quenched fluorescence

SA

Fl

FlFl

Fl

SA

Enhanced fluorescence

Assay Set-up

▪ PBS, PBS+0.1%BSA, and PBS+0.1% glycerol

▪ Add 25 µL of biotin-4-fluorescein (Fl) to black 96w

plate

▪ Add 25 µL of inhibitor

▪ Add 25 µL of streptavidin (SA)

▪ Incubate for 30 min at room temperature

▪ Read fluorescence: Ex = 485 nm and Em = 515 nm

© 2019 Artel20

© 2013 ARTEL

Assay Development Parameters Evaluated

Parameter Manual Automated

Solution stability X X

Plate type X X

Background buffer X X

Inhibitor X X

Inhibitor conc. X X

Incubation time X X

Mixing X X

Pipette calibration X

Asp/Disp rates X

Fluid exit rate X

Air gaps X

© 2019 Artel21

© 2013 ARTEL

PB

S M

an

ual

PB

S A

uto

BS

A M

an

ual

BS

A A

uto

GL

Man

ual

GL

Au

to

PB

S M

an

ual

PB

S A

uto

BS

A M

an

ual

BS

A A

uto

GL

Man

ual

GL

Au

to

0 .0

0 .2

0 .4

0 .6

IC5

0 (

M)

N o M ix

M ix

Effect of Mixing on Assay Variability

Key Takeaway

Buffer has an impact on

mixing; Mixing improved BSA

and glycerol buffers

PBS = phosphate buffered saline / BSA = PBS + 0.1% bovine serum albumin / GL = PBS + 0.1% glycerol

Mixing was conducted by

aspirating and dispensing 3

cycles after the third reagent was

added. The plates were then

incubated for 30 minutes at room

temperature.

© 2019 Artel22

© 2013 ARTEL

-2 -1 0 1 20

1000

2000

3000

4000

Log Inhibitor Concentration (nM)

Sig

na

l (

RF

U)

PBS

BSA

Glycerol

Effect of Buffer Type on Assay Performance

-2 -1 0 1 20

1000

2000

3000

4000

Log Inhibitor Concentration (nM)

Sig

na

l (

RF

U)

PBS

BSA

Glycerol

Manual Mixing Auto Mixing

Key Takeaway

Manual mixing can behave differently than automated

mixing, especially depending on assay buffer ingredients.

© 2019 Artel23

© 2013 ARTEL

Effect of Source Plates on Automated Liquid Handler Verification

A

B Key Takeaway

Source plate affected the destination

plate – observed by visual inspection

Artel MVS plates containing

reagent from non-binding (A) and

untreated (B) 96w black source

plates.

MVS precision: 6.3% for non-

binding and 1.2% for untreated

© 2019 Artel24

© 2013 ARTEL

PBS M

anual

PBS A

uto

GL M

anual

GL A

uto

PBS M

anual

PBS A

uto

GL M

anual

GL A

uto0.0

0.1

0.2

0.3

0.4IC

50

(

M)

Non-binding

Untreated

Effect of Plate Type on Assay Variability

Key Takeaway

When subjected to the assay, the

uncoated plate yielded slightly

lower variability.

PBS = phosphate buffered saline / BSA = PBS + 0.1% bovine serum albumin / GL = PBS + 0.1% glycerol

Black, 96-well plates used:

• Corning #3650 (Non-binding)

• Corning #3915 (Untreated)

© 2019 Artel25

© 2013 ARTEL

Effect of Reagent Temperature on Assay Variability

-2 -1 0 1 20

1000

2000

3000

4000

5000

Log Inhibitor Concentration (nM)

Sig

nal (R

FU

)

Cold Reagent Dispense

Room Temp Dispense

IC50 (µM)

0.311

0.235

Key Takeaway

Cold reagent dispense

yielded slightly lower potency

“Cold reagents” were stored at 4°C until use. The reagents were not maintained at 4°C during pipetting.

© 2019 Artel26

© 2013 ARTEL

Effect of Aspirate (ASP) Speed on Assay Performance

-2 -1 0 1 20

1000

2000

3000

4000

Log Inhibitor Concentration (nM)

Sig

nal (R

FU

)

Aspirate 1 PBS

Aspirate 5 PBS

Aspirate 5 GL

Aspirate 1 GL

Aspirate 1 BSA

Aspirate 5 BSA

IC50 (µM)

0.157

0.163

0.151

0.225

0.145

0.207

Key Takeaway

Aspirate speed affected potency for

glycerol and BSA buffers

PBS = phosphate buffered saline / BSA = PBS + 0.1% bovine serum albumin / GL = PBS + 0.1% glycerol

© 2019 Artel27

© 2013 ARTEL

-2 -1 0 1 2

0

1000

2000

3000

4000

Log Inhibitor Concentration (nM)

Sig

nal (R

FU

)

Dispense 1 BSA

Dispense 5 BSA

Dispense 1 GL

Dispense 5 GL

Dispense 1 PBS

Dispense 5 PBS

Effect of Dispense (DISP) Speed on Assay Performance

IC50 (µM)

0.268

0.2660.136

0.197

0.164

0.167

Key Takeaway

Dispense speed affected potency

for glycerol but not PBS or BSA

PBS = phosphate buffered saline / BSA = PBS + 0.1% bovine serum albumin / GL = PBS + 0.1% glycerol

© 2019 Artel28

© 2013 ARTEL

-2 -1 0 1 2

0

1000

2000

3000

4000

Log Inhibitor Concentration (nM)

Sig

nal (R

FU

)

AirGap0BSA

AirGap2BSA

AirGap5BSA

AirGap0GL

AirGap2GL

AirGap5GL

AirGap0PBS

AirGap2PBS

AirGap5PBS

Effect of Air Gap on Inhibitor Potency

IC50 (µM)

0.243

0.269

0.171

0.179

0.155

0.163

0.172

0.175

0.205

Key Takeaway

Air gap affects the assay containing PBS

and BSA buffers with respect to potency and

variability

PBS = phosphate buffered saline / BSA = PBS + 0.1% bovine serum albumin / GL = PBS + 0.1% glycerol

© 2019 Artel29

© 2013 ARTEL

PB

S

BS

A

GL

PB

S

BS

A

GL

PB

S

BS

A

GL

0 .0

0 .1

0 .2

0 .3

0 .4

IC5

0 (

M)

A irG a p 0

A irG a p 2

A irG a p 5

Effect of Air Gap on Assay Variability

PBS = phosphate buffered saline / BSA = PBS + 0.1% bovine serum albumin / GL = PBS + 0.1% glycerol

© 2019 Artel30

© 2013 ARTEL

Study Summary

▪ Four assay parameters were identified which required optimization both

during development at bench-scale and on the ALH.

▪ Assay buffer selection

▪ Mixing

▪ Reagent stability

▪ Plate type

▪ Certain ALH parameters were dependent on buffer type. Not all ALH

parameters will effect an assay.

© 2019 Artel31

© 2013 ARTEL

Conclusions

▪ Performing assay optimization AND liquid handler optimization together as

a whole process reduces potential for error introduction and

prolonged/difficult method transfer.

▪ Evaluate critical liquid handling parameters and potential sources of

variability at bench scale and on the ALH for each new assay.

Assay optimization or LH qualification alone isn’t enough.

© 2019 Artel32

© 2013 ARTEL

Putting It

Together… Assays

depend on

reagent

concentrations

Reagent

concentrations

are volume-

dependent THEREFORE:

Assay

integrity is

dependent on

accurate

volume

deliveryAssay results are impacted by

liquid handling variability

© 2019 Artel33

© 2013 ARTEL