HIGH CONTENTSCREENING

High content screening (HCS) is a remarkabletechnology that allows fully automated cellularimaging on a truly high-throughput scale. In thelife and biomedical sciences it is now widelyemployed to solve questions in fundamentalscience, providing the research community withnovel views on how cells function. In parallel,HCS is also playing an increasing role in thedesign and development of new therapeutics,providing key information to inform drugdesign and delivery. This article provides anintroduction to this technology, and perspectiveon its future direction.

The technology of high contentscreening

The basic unit of all organisms is the cell.Ultimately it is cells that are responsible for allof our daily activities – walking, talking,visualising the world around us, intelligentthought. Therefore, gaining a greaterunderstanding the cell and its regulation, andhow each cell interacts with its neighbours intissues and organs remains a fundamental goalfor life scientists in the wider context of healthand disease. While physiologists tend toconsider this at the organism or tissue level,biochemists and cell biologists generally preferto work with individual cells, on the premise

that if they can understand the molecularcomplexity of a single cell growing in culture inthe laboratory, then the information obtainedcan potentially be extrapolated to different celltypes and perhaps the entire organism.

A landmark event for life scientists at the turn ofthe century was the sequencing of the humangenome. Having access to the detail of the codeof life provides us with the ‘raw information’ to understand how cells function. The 22,000human genes encode proteins, each of whichhas a defined cellular function and a discretenetwork in which it operates. The challenge ofcourse is how to decipher this network, andassign each protein to particular functions inthe cell. Furthermore, there is also the need tobe able to identify those proteins that make ussusceptible to infection, or that act as driversfor specific diseases. For more than 350 years,microscopy has arguably been the primary toolfor biologists to study cells and attempt to revealhow they function, although its application tostudy individual molecules was limited. In thelatter half of the twentieth century a number ofkey limitations were overcome, principallythrough technologies allowing the introductionof fluorescent probes into cells, in turn providinga renaissance to microscopy. For the first time itbecame possible to rapidly and specificallyhighlight particular proteins, and visualise dynamicprocesses in live cells. Effectively, fluorescence

High content screening:A technology impacting both fundamental andtranslational science in the twenty-first century

Prof. Jeremy C. SimpsonCell Screening LaboratoryUniversity College Dublin (UCD)Email: jeremy.simpson@ucd.ie

2

microscopy became more sophisticated, and inparallel increasingly quantitative. Fluorescenceimaging of cells has revealed the intricate detailof their architecture and the spatial and temporalarrangements of proteins encoded by the genome.However, microscopy was then faced with anew challenge, namely how to image on a trulyhigh-throughput scale, yet still preserving imagequality and the depth of information containedwithin. The automated microscopy that evolvedfrom this conundrum is termed high contentscreening microscopy and analysis – HCS/HCA.This is a fusion technology, combining laboratoryautomation with sophisticated software routinescapable of analysing images of millions ofindividual cells and extracting used-definedquantitative information – ‘content’ – for eachcell. The real power of this approach is that it isnow no longer essential to manually examine

and curate each microscopy image, which inturn has the beneficial consequence of removinghuman bias.

HCS can in principle be applied to study anycell-based biological question. At its heart istypically an assay in multi-well plate format,designed with one or several fluorescent reportersthat may change their intensity or distribution inthe cell during the course of the experiment.Ultimately any changes must be detectable andbe able to be quantified in a reliable manner.Another key facet of the technology is that asmuch of the experimental pipeline as possibleshould be automated. Laboratory automationhas evolved massively in the last ten years, withsmall-scale and large-scale solutions available tosuit both individual research laboratory groupsand multi-national pharmaceutical companies.

Number of samples

Experimental detail obtained

High content screening microscopy provides the capacity to image a largenumber of samples, and preserve the high level of detail within each

High content screening bridges the gap between image quality and quantity

3

Automation can be at the level of liquid handling,cell preparation, reagent dispensing, and ofcourse the automated imaging and analysis; butin each case strict quality control is required toensure that each sample of series assays isprocessed in a highly parallel manner. HCSassays deliver image data on a massive scale,with individual screening campaigns producingdata that run into the tens of terabytes. At the coreof the technology is the analysis component –not only does every image need to be analysed,but indeed every cell in every image needsindividual quantification, thereby preserving the‘content’. Early HCS experiments tended to uselow power microscope objectives, as typicallythe readouts from the assays were relativelysimple, for example total fluorescence intensityof a marker in question, and as such the spatialdetail was less important. However morerecently HCS experiments have becomeincreasingly sophisticated, with advancedsoftware routines measuring intricate detail offluorescence distribution in cells and indeed inindividual cellular structures. Analysis routinescan be combined with machine learning toautomatically classify cell phenotypes intodefined populations. Furthermore, when theanalysis software is running in real-time alongsidethe image acquisition it is also possible for theautomated microscope to return to a particularwell position to acquire further high-resolutionimages of particular phenotypes identified in thecells. HCS technology and the image analysispossibilities continue to evolve at a rapid rate,and in a way are only limited by the creativityof assay design.

Applying HCS in fundamentalscientific research

It is clear that HCS can be applied to address ahuge range of fundamental questions about cellThe high content screening pipeline

4

behaviour and function. As mentioned above,mammalian cells display a complex internalarchitecture, with one fundamental premisebeing that all the cells’ components operate indistinct locations with specific distributions.Certain material also periodically needs to movefrom one location to another (including betweenthe inside and outside of the cell), and anydefects in this can have serious consequencesfor the cell. These observations all provideopportunities for assays to be designed tomonitor change. Similarly most cells at steadystate have a defined size and shape, featuresthat also lend themselves well to automatedanalysis and quantification. Using fluorescentreporters, typical cellular parameters that HCSassays measure can broadly be divided intothree categories:

• morphological (area, width, length, circularity)

• intensity-based (number, average, distribution,variation, co-localisation)

• texture (granularity, topography).

Many successful HCS campaigns havetherefore utilised basic cell biology knowledge,combined with these analysis possibilities todesign assays to identify proteins and networksregulating particular cell functions. Examplesinclude the protein networks required tocontrol cell division, migration, signalling, theinternalisation or secretion of molecules, andcell death, to name but a few. Other excitingmicroscopy screens have examined themechanisms and pathways used by bacteriaand viruses to enter target cells and initiateinfection. All of these examples can beclassified as either basic cell functions, orchallenges that cells are likely to encounter,and so HCS presents itself as a marvelloustechnological tool to systematically probe howsuch events are regulated. Although thisrepresents fundamental scientific research, it isinevitable that the findings from these screenswill ultimately have consequences with respectto our wider understanding of cell health. Asmore proteins (and their dysfunction) becomelinked to disease, it is also clear that these datawill inform possible interventions.

Raw image Image segmentation& data extraction

Data visualisation& interpretation

Cell area: 33751 Nucleus Area: 8000Nucleus Roundness: 0.82

Texture feature 1: 0.548

Features

Cell area

Cel

ls

No.

of c

ells

Organelle area: 2720

EGFP max. Int. 4035

High content screening facilitates automated image analysis and easy data visualisation

5

Potential impacts of HCS intranslational science

Interestingly HCS has its roots in thepharmaceutical, rather than the academic sector.For many years the identification of noveldrugs and their targets was hindered by thelack of information coming from traditionalhigh-throughput screening of compoundlibraries, largely because they employed simplecolorimetric assays which were unable to revealsub-populations of cells within a single well of amulti-well plate producing differing responsesor phenotypes. Microscopy-based imaging ofeach well was the obvious solution to thisproblem, and hence automated imaging andanalysis of cells was developed and becameestablished technology. In the pharmaceuticalsector it is now clear that even subtle differencesbetween how cells respond to compounds iscritical information that can be used to informdesign and dose of the potential therapeutic.

HCS has also found a role for itself in pre-clinicaltoxicity studies. This is an important developmentbecause cell-based assays can reveal a number oftoxicity-related phenotypes in cells – for examplechanges in the shape of the nucleus, alterations inmitochondrial potential, permeability of the cellmembrane – all in parallel in a single experiment(so-called ‘multiplexing’). The application ofHCS to toxicity studies not only enhances theinformation that we have about a potentialtherapeutic agent, but its use also significantlyreduces the amount of animal testing required.Another emerging translational application ofHCS is in the area of drug delivery. It is becomingclear that even after the development of a newdrug, its ability to be targeted correctly is notguaranteed. HCS is playing a role in screeningfor drug carriers and formulations that provideoptimal targeting and delivery to particularcell types, again information that is incrediblyvaluable for the smarter design of therapeutics.

Outlook

The ever increasing role that HCS technologyplays in the life sciences is impressive, howeverit is a technology that must continue to evolve tochanging needs and the increasing demands ofresearchers. One area that is seen as particularlychallenging is in the use of more complexthree-dimensional cell culture models. Thepower of HCS lies in its ability to extract detailedquantitative information from single cells,information that accurately portrays the stateof that cell. One criticism however is that themajority of cell types used in HCS experimentsare effectively two-dimensional, and do notsignificantly resemble cells found in theorganism from which they were originallyderived. It is now relatively straightforward toculture clusters of cells in three dimensions, forexample as ‘cysts’ or ‘spheroids’, but theysuffer from the problem that each cell in the cystis small, and therefore the quality of spatialinformation that can be extracted is poor. Suchcell models will undoubtedly add extra value,particularly in pharmaceutical research, butimaging and image analysis difficulties willneed to be overcome.

Another area that HCS needs to address is itsstandardisation. A wide variety of hardwareand software solutions are available, but to datethere have been few attempts to standardiseimage types and image analysis algorithmsand protocols across different platforms. This would be highly beneficial, as it wouldenable more efficient data mining from thelarge image sets publically available, and as aconsequence the extraction of even more‘content’ from the images. Nevertheless, HCSappears to be a technology that is nowfundamental to so many strands of biologicalresearch, and for the foreseeable future it ishere to stay as part of the search for answers tothe big questions in the life sciences. ■

6

One of the most remarkable things about life onearth, in all its forms, is how cells often only tensof microns in diameter have evolved to carry outthe variety of tasks that they do. In multicellularorganisms the situation is even more complicated,as different cell types need to work together inan orchestrated manner in functional units suchas tissues and organs to maintain the health ofthe organism. Ultimately therefore, it is thefunction of individual cells in our body thatdetermines our health, and our susceptibility to

disease and infection. The discipline of cellbiology serves to understand how cells work,and importantly what goes wrong in cells tocause disease. It is a discipline, with associatedtechnologies, positioned at the centre of allfundamental biomedical research.

Since the mid-seventeenth century microscopyhas been the primary tool for scientists to revealthe structure and organisation of cells, both in isolation and in their ‘social’ context. In the late

It’s the ‘content’ of cells thatmatters in biomedical research

7

twentieth and early twenty-first centuries however,the widespread application and integration offluorescence technologies with microscopy hasprovided new opportunities to reveal theinnermost workings of cells. Fluorescencemicroscopy allows researchers to potentiallyview not only the cellular organelles, but also thebillions of molecules – in particular proteins –that work together to provide the cell with itsfunctionality. Therefore, in this post-genomesequencing age, how can we assign discretefunction to each of the 22,000 human genes andthe proteins that they encode? Furthermore,how can we identify those proteins that cancause particular disease, and those proteins thatcan have protective properties? Carrying outsuch experiments in intact and preferably livingcells has obvious benefits, but clearly the scaleof such experiments is challenging. Simplyvisualising each protein in turn (and moleculartechniques in principle make this possible)requires 22,000 individual microscopy images,and so without considering any furthercomplexity of the experiment or replicates wewould need to image almost every well from230 96-well plates in a consistent manner. Even ifthis is achievable, the next problem becomesone of how to interpret the images, and in sucha way that we can objectively compare them.The issues are experimental scale and complexityof information.

In the last ten years this experimental approachhas become a reality, and these barriers are beingovercome, encompassed in a technology termed‘high content screening and analysis’ – HCS /HCA. This is a fusion technology, combininglab automation, particularly in terms of themicroscopy, with sophisticated softwareroutines capable of analysing images of millionsof individual cells (‘high throughput’) andextracting user-defined quantitative information(‘content’) for each cell. Since its development,

labs around the world have embraced its powerto address both fundamental cell biologyquestions and applications relevant to humanhealth and disease. HCS can and has been usedto rapidly screen massive libraries of chemicalcompounds to identify leads with desiredcellular phenotypes, to identify host factorsassociated with virus infection, and reveal newtriggers for cancer cell development. For cellbiologists it has proved to be a particularlypowerful technology when combined with RNAinterference (RNAi), a molecular technique thatallows researchers to inactivate genes and theproteins that they encode in a systematicmanner. Carrying out RNAi experiments in anHCS format effectively allows us to dissect thefunction of each gene / protein in turn withrespect to a particular biological question, withthe output being images of cells revealing thephenotype, and also their quantitative analysis.

In the Cell Screening Lab at UCD (www.ucd.ie/hcs)we have been developing and applying HCSstrategies for a number of years to addressquestions related to how cells transport material(cargo) between their various internal organelles.Understanding how these membrane transportprocesses work is of vital importance, as allcargo inside cells will only facilitate cell functionif it is located in the correct place. For example,signalling receptors at the cell surface areactually synthesised and assembled in internalmembranes of the endoplasmic reticulum,requiring transport through intermediateorganelles for further processing prior to deliveryto the cell surface. Many human diseases areassociated with mistargeting of such receptors –cystic fibrosis is a well-known example. Ourongoing mission is to use RNAi at a wholegenome scale to systematically dissect how suchtransport pathways are regulated, and ultimatelyto use this information to gain insight into howthey can be manipulated. Improving drug

8

delivery efficacy into cells is one good exampleof how this approach can be utilised. Ultimatelytherefore, we believe that our HCS approachesprovide critical information about cellorganisation that can be exploited by manybranches of biomedical research.

There are of course both technological andpolitical challenges to overcome if HCS is tocontinue providing valuable data to the scientificcommunity. From a technological perspectivethere is a move towards use of more complex 3-dimensional multi-cell type models, whichalthough may better represent the in vivosituation, they are more difficult to image andprecisely quantify, requiring confocal HCStechnology. Politically, HCS is a relativelyexpensive technique, both in terms of itshardware and the reagents needed to carry outlarge-scale screens. In the current challengingenvironment of research grant availability,

funding bodies often prioritise more advancedor applied projects that might return gain inthe short term. Ignoring HCS projects would befoolish, as they show real promise to deliveradvanced cell biology knowledge that will informand drive the direction of future biomedicalresearch. ■

9

Left: image of two cultured human cells fluorescently labelled for various subcellular structures. Right: example of ‘content’ that can be extracted from these cells

CELL 1 : INTENSITYCHANNEL 2 : COMPOUND A1567MEAN : 1025MAX : 3267MIN : 10SUM : 145002CONTRAST : 23565

CELL 1 : MORPHOLOGYLENGTH : 338.117pxWIDTH : 218.257pxAREA : 59280px2

ROUNDNESS : 0.44PERIMETER : 1034px

CELL 2 : SPOT CHARACTERISATIONAREA : 2130px2

ROUNDNESS : 0.83INTENSITY : 2031NUMBER : 56

Prof. Jeremy C. SimpsonCell Screening LaboratoryUniversity College Dublin (UCD)Email: jeremy.simpson@ucd.ie

Prof. Jeremy C. SimpsonCell Screening Laboratory

University College Dublin (UCD)Email: jeremy.simpson@ucd.ie