How to Develop an HTS Compatible Assay

How to Develop an HTS Compatible Assay

Lucile White

Southern Research Institute

High Throughput Screening Center

What is HTS?

It is a system that uses specialized automation equipment and high density microtiter plates to screen a large number of “wells” in a short period of time. Throughput of 30,000 to 100,000 compounds per day is common.

Key System Components

Compound management Precision robotics for liquid and plate handlingInformatics Associating data with a particular compoundAnalyzing data from a screenCheminformaticsPeople

Since

2006,

screened over 3 million compounds each year;

3.5

million in

2009

In 30 – 40 assays/year

Percent of Compounds Screened by Assay Type

Percent Compounds Screened by Biocontainment Level

Neils Bohr’s Definition of an Expert

An expert is someone who has made every mistake possible within

a

very narrow field of inquiry

The HTS Center’s role is to help you not make the same mistakes we have already made

Why is this a

bottleneck?

Assay

Development

When you go from this:

To this:

THE RULES CHANGE

What are you aiming for in an HTS assay

To have

a

reasonable chance to believe the results of

a

single determination, i.e. one well

For that you need

Reproducibility from well to well

Reproducibility from assay plate to assay plate

Reproducibility from day to day

Rules Changes

Mode of Detection

Mode of Manipulation

Reproducibility in prep

No Physical Intervention

No Protocol Changes – No if/then scenarios

Counter Screens and Secondary Assays

Cost for Failure

Mode of Detection

Detection methods simplified (i.e. homogenous mix and read)

Plate readers (not high content)

Fluorescence

Luminescence

Absorbance

Fluorescence Polarization

FRET; TR-FRET

AlphaScreen

(Perkin Elmer)

RT-PCR

siRNA

Not available at SRI

Radioactivity

Formats that need to be read within seconds after addition of reagent

Flash luminescence

FLIPR – Calcium channel sensing

Why Not HCS

Expensive and time consuming especially when 90-99% of the compounds will have no effect

To run as HTS, cells must be fixed

Fixing and adding dyes and/or antibodies require wash steps

Clearly some types of questions can only be answered by HCS; however

at this time

I do not recommend this approach for HTS

Currently useful for 2

nd

or 3

rd

level assays – low to medium throughput – known actives

New Technologies at SRI2010

RT-PCR

For HTS it needs to be compatible with 1536-well plates

The 1536 instrument can process

a

plate in less than an hour allowing high throughput

qPCR

for screening operations

Not the best option because of high cost

Available by end of 2010

siRNA

(collaboration with Dr. Bjornsti)

Ambion

siRNA

human genome library

21,585 human targets with 3 non-overlapping

siRNAs

for each target; Silencer Select v4

Available by end of 2010

Infectious Diseases

Bacterial Motility – soft agar swarm assay

Bacterial Biofilm

Antivirals- developing new methodology where CPE is not required

Mode of Manipulation

Robotic

Shakers

Non manual dissociation (can not do it with your hand, fingers, scrapers, etc)

In general no centrifugation

In general no separation steps

In general no wash steps

If it can't be done by pipetting or shaking probably not HTS

Reproducibility in Prep

Every steps adds variability

Minimize variation in steps (

eg

. dispensing cell suspension – minimize clumping)

Minimize chance of error (eg. Ease of pipetting solution)

Minimizing variability may not mean choosing conditions that give the maximum signal but optimizing around parameters where small changes produce very little change in the signal

Provide excess time for step when available (eg. lyse for 20 minutes when 5 minutes is usually sufficient)

Use stable cell lines rather than transient transfections

No Physical Intervention or Protocol Changes

Robots can’t make judgment call (i.e. it looks like the cells are lysed, it looks like the solution is mixed)

Can’t let cells grow another day

Can’t incubate longer for an enzyme assay

Ensure solutions pipette without clogging

Other Issues

All HTS assays are subject to some type of “compound” interference

Interference increases with lower wavelengths

Red shifted dyes preferred, eg GFP not good endpoint reagent

Time-resolved delays help

Ratiometric

methods have

not

been useful in our hands

Incubation temperature – room temp

Cost associated with thousands of hits

Average hit rates 0.5-3% so sooner you identify what is real and what is not the better off you are.

The Rules of HTS

Begin with the end in mind (Stephen Covey)

Design an HTS assay from the beginning

Don’t try to automate

a

manual assay after the fact

KISS- Keep It Simple Stupid (Kelly

Johmson

)

Simple robust assays are best

Fewer variables increase the chance of success

Two General Types of Assays

Drug discovery assays can be divided into two broad classes: biochemical and cell-based

Both assay formats are amenable to miniaturization and adaptation to high-throughput screening

Biochemical Assays:pHTemperature (room temperature)Ion ConcentrationReagent StabilityReagent AggregationReagent SolubilityOrder of Reagent AdditionSolvent Effects (DMSO)Reagent Concentration

Some Causes of Assay Variation

Cell and Organism Based Assays As for Biochemical assays, plus: Cell culture plastics Culture media Culture conditions Serum Cell cycle Passage number Solvent effects (DMSO) InfectionTransient transfectionsHeterogenous population of cells

Some Causes of Assay Variation

Statistical Analysis: Z Factor

3SD of sample + 3SD of control

mean of sample – mean of control

Z = 1 –

Zhang, Chung and

Oldenburgh

, J.

Biomol

. Screening,

4:67-73,

1999

Separation

Band

Data variability

Band

Data variability

Band

m

s

m

c

Frequency

Assay Signal

3s

s

3s

c

Statistical Analysis

Z-plate analysis defines edge and row effects as minimalLiquid handling methods perform as expectedZ = 0.86S/B = 83S/N = 24 Excellent Data!

What You Assay is What You Get

Capture the right biology

Balance with making assay so complex it can not be automated

Every assay has artifacts in the form of false positives and false negatives.

Design and use counter screens and secondary assays

There is such a thing as a too stringent assay.

Very high Z-factors can often indicate that the bar for something to show up as an active is set very high.

Is the assay still in the linear range?

Costs for Failure

$20-40,000 for every failed day in HTS

HTS schedule usually finalized a week to a month in advance – failures cause a negative ripple effect in rearranging the schedule

Useful References for HTS Assay Guidance

Assay guidance manual

www.ncgc.nih.gov/manual/toc.html

Joint effort of Lilly and NIH scientists

Sections on assay development issues for specific assay formats

Reporting data from HTS/information I am looking for to review an ADDA proposal

Inglese et al, Nature Chemical Biology,

3

:438, 2007.

Troubleshooting cell based assays

Maddox et al, J. Assoc. Lab. Automation

13:

168-73, 2008.

Compound Libraries

Which library we will use does not need to be specified for the ADDA proposal

We will determine that in conjunction with the SRI chemists once we are closer to performing the screen

Southern Research has purchased several libraries from commercial suppliers for use in its HTS program

One example,

a

100,000 compound library from ChemBridge Corporation was pre-filtered using

a

number of drug-likeness parameters including molecular weight <500, <5 hydrogen bond donors, and <10 rotatable bonds. Over 28 reactive groups such as

carbodiimides

, diazonium salts, peroxides, and diazines were also removed, resulting in 100,000 compounds with drug-like properties that meet

a

number of criteria including the Lipinski rules.

Specialty libraries

Kinase library

NIH Small Molecule Repository

The Three Cardinal Rules

in

Developing

an HTS Compatible

Assay

Do Everything You Can to Minimize Variation

Well to Well

Plate to Plate

Day to Day

Do Everything You Can to Minimize the Chances of a Catastrophic Failure

HTS has the Flexibility of a Mac-Truck

So unless you have an extra 250k to pay for us to make the changes, it’s best to conform to our process

Properly Designed

Poorly

Designed