Laser in Cytometry Applications in Flow Cytogenetics C Cremer J Doumllle M Hausmann F F Bier

Institut fuumlr Angewandte Physik I Universitaumlt Heidelberg Albert-Ueberle-Straszlige 3 - 5 D-69oo Heidelberg Federal Republic of Germany

P Rohwer

Institut fuumlr klinische Immunologie und Rheumatologie Universitaumlt Erlangen Krankenhausstraszlige 12 D-8520 Erlangen Federal Republic of Germany

Chromosome Analysis Flow Cytometry I Fluorescence I Laser Applications I Methods and Systems

Flow cytometry has become an important tool in biomedical research and elinical diagnosis In particular its ability to make multiple measurements of individual particles at a rate of up to 20000 particles per second allows the quantitative characterization (flow karyotyping) and sorting of isolated chromosomes Major areas of application of flow cytogenetics are in molecular genetics human genetics and tumor cytogeneuumlcs Several approach es to increase the resolution of one and two parameter flow cytometry of chromosomes wil1 be discussed On one hand different algorithms of flow cytometric data processing can be applied to achieve a better signaljapparent noise ratio To obtain a more accurate evaluation of flow karyotypes they may be approximated as a sum of Gaussian distributions On the other hand the flow cytometric resolution of specific chromosome types can be improved by chemical modifications of the chromosomal

DNA eg incorporation of base analogues or in situ hybridization in suspension with specific DNA probes

Perspectives and Overview

Lasers have found a variety of significant applications in biomedical sciences They are used in manyfold ways to obtain quantitative information on cells cell organells and parts of cells such as chromosomes This report will be limshyited to a special application being of growing importance in many branches of biological and medical sciences the use of lasers in flow cytometry

In conventional flow cytometry (also called zero resoshylution flow cytometry because the dimension of the morshyphological resolution is zero) one or several fluorescence values resulting from excitation with continuous-wave (CW) laser light are measured for each partic1e stained with a specific fluorochrome According to the variety of fluoresshycent staining procedures characterization of biological parshytic1es is possible in many different ways e g DNA content DNA base ratio internal substrate concentrations surface antigens etc In addition some gross morphologie features

may be extracted by evaluation of forward (05deg - 20deg) and right angle scattered laser light (eg information on size and shape)

The power offlow cytometry (as compared to quantitative fluorescence microscopy) derives from the ability to make multiple measurements of each partic1e simultaneously at very rapid rates Presently a rate of several thousand parshytic1es per second may be achieved in commercially available flow cytometers in specially designed instruments [1] the rate may be increased up to 20000 particles per second

Another important advantage is that partic1es may physshyically be separated due to preselected optical parameters Under favourable conditions high purities (up to 98) may be 0 btained

Duc to its quantitative capacity versatility sensitivity speed and ability to identify and sort cell subsets in hetershyogeneous populations flow cytometry is presently used world wide in hundreds of biological and biophysical re-

Ber Bunsenges Phys ehern 93 327-335 (1989) copy VCH Verlagsgesellschaft mbH D-6940 Weinheim 1989 0005-9021890303-0327 $ 02500

C Cremer ct aL Laser in trllTlptr AP1PllcaUC)ns in Flow328

search laboratories and medical institutions For example flow cytometric measuretnents of cellular DNA content alshylow the quantitative characterization of a variety of tumor cell lines and thus have found a vast application in tumor diagnosis [2 3J

This report will focus on some aspects of the use of flow cytometry for quantitative characterization of chromosomes rltflow cytogenetics) Since a general review on this field has been published recently [4] this artic1e in particular will discuss in more detail some approaches to increase the resshyolution of flow cytometric data

Principles of Laser Fluorescence Acthated Analysis and Sorting

The technique of laser fluorescence activated analysis and sorting of cells and chromosomes has mainly been developed in the last two decades [5J and has become important for many applications both in biomedical research and c1inical diagnosis (for general reviews see [6 - 9J)

Schematic design of a single laser beam flow cytometer and sorter See text for details (by courtesy of Coulter Corp)

Its least complex form is presented in Fig 1 A suspension of partic1es to be measured for instance individual cells or isolated chromosomes is injected coaxially into the sheath fluid stream of the flow chamber (flow celI) By means of hydrodynamic focusing through a jewel orifice of 50- 200 Jlm diameter (typical diameter for chromosome analysis 70- 80 JLm) at the exit of the flow cell the particles are confined within the core of the stream of only a few micrometers diameter (typically about 10 of the jet diashy

passing one by one a slightly focused CW laser beam Tbe high flux excites the particles stained with fluoroshychromes specific for example to DNA DNA base ratio or cell surface antigens

In most cases argon- and krypton-ion lasers (excitation of dyes in the visible and near UV) are used with total emission power of several watts For some immunological applications helium-neon or ion-pumped dye lasers are used too A high stability is required for precise measureshyments because for small partic1es as chromosomes any varshyiation in laser power may result in a variation of the emitted fluorescence intensity

For eaeh partic1e the fluoreseence emission is deteeted by photomultipliers (PMT) during the few microseeonds taken to traverse the foeal volume The PMT signal is either linshyeally or logarithmieally integrated digitized and registered by a multi-ehannel analyzer The larger for instance the DNA content of a cell nuc1eus or a chromosome is the more dye may be bound and consequently the fluorescence intenshysity inereases Thus it is possible to discriminate different types of partic1es aceording to their fluorescence properties

Using a single laser be am device two fluoreseenee parashymeters (following simultaneous staining with two fluoroshychromes of similar absorption spectra) may be measured if the fluorescence emission speetra of the two dyes do not overlap In this ca se e g green and red fluoreseenee are separated by appropriate edge or band pass filters In adshydition the laser light scattered by a partic1e mayaiso be measured and used for further diserimination For this forshyward and right angle light seatter is detectable Nowadays many commercial machines are equipped with more than four detectors and two lasers for two three or multicolor fluorescenee measurements

For physical separation of specific types of particles (sorshyting) a hydrodynamic electronie procedure is applied which allows to sort them into collection vessels aecording to preshyseleeted optieal properties measured as deseribed above To do this the fluid jet in air is disrupted into single droplets (diameter about 18 of the jet diameter) after having passed the laser intersection point This is realized by mechanical vibrations of the flow cell up to 40 kHz in vertical direction parallel to the streaming jet Those droplets eontaining parshyticles to be sorted are charged electrically before separation from the jet and deflected into recipient vials by an eleetroshystatic field of a few kVjem As mentioned above the method presently allows analysis and sorting of particles between several hundreds and 20000 per second

Flow Cytometry of Chromosomes Univariate Distributions

In this report we shall foeus on a special feature of laser fluorescenee activated analysis and sorting i e flow cytoshymetry of isolated metaphase chromosomes in suspension (also called flow cytogeneties) ([10J for review see [4J) This has found important applications in moleeular biology human genetics and c1inical diagnostics of genetic diseases Presently one of the main applieations of chromosome sortshying is the construction of chromosome specific DNA librarshyies [11 12J

Fig 2 gives an exampIe of a so called univariate flow karyotype using one fluoreseenee parameter only On the ordinate the number of the fluorescence signals (partic1es) is given on the abscissa the logarithm of the fluorescence inshytensity is plotted Ideally eaeh fluorescence signal represents

I

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 329

Imiddot a

(

3M I

200

192

96

8 -1 161 242

Fig2 U nivariate flow karyotype of chromosomes from the Chinese hamshyster cell line CHL (DAPI staining laser excitation in the near UV) ordinate Number of particles abscissa Channel number (in this case corresponding to the logarithm of the fluorescence intensity) The peak designation (No 1 - 9) was made according to decreasing fluorescence intensity (correlated with the chromosome length in case of chromosomes of the Chinese hamster) The flow karyotype shown has been obtained without further processing of the data

one chromosome with a specific DNA conte nt In the case shown in Fig 2 chromosomes isolated from a Chinese hamshyster cell line were used The chromosome suspension was stained with the fluorochrome DAPI (5 JlM) specific for doublestranded DNA Excitation was done by an argon-ion laser (UV multiline 3511 - 3638 nm) with 50 mW Under favourable circumstances and optimum resolution each chromosome type of a cell line is represented by one indishyvidual peak One possibility to assign the peaks to individual chromosome types is to sort the peak contents and to idenshytify the chromosomes by conventional cytogenetic methods [1013] Another preliminary assignment is usually possible by comparison of the flow cytogenetic data with light mishycroscopic observations Under appropriate conditions ie that the ATGC correlation is about the same for each chroshymosome type an excellent correlation may be obtained beshytween chromosome length and relative fluorescence intensity of the flow karyotype

To achieve the flow karyotype shown in Fig2 a chroshymosome isolation method [14 15J was applied using a hexshyandiol isolation buffer to stabilize the chromosomes This procedure has the particular advantage that chromosomes in suspension may be stored at 4degC for long times (in our experience up to 5 years) without a significant deterioration of the resolution of the flow karyotype [16]

Data Evaluation

In many cases the flow karyotypes obtained under various experimental circumstances are not as weIl resolved as shown in Fig 2 This may be caused by different preparation conditions one of them is the buffer system of the chroshymosome suspension which may not always be optimal for flow karyotyping but necessary with respect to other expershyimental studies done with the chromosomes

For example free flow electrophoresis of isolated metashyphase chromosomes requires a buffer system of low ionic

strength [48] Flow cytometric characterization mayJeadi~~ to poorIy resolved flow karyotypes because of morpbolo~~r~ ical instabilities of the chromosomes in this buffer ili~fl

In Fig 3a the experimental flow karyotype of chromc)~i ~ somes of a Chinese hamster cell line in free flow electrb~ phoresis buffer prior to data processing is presented~Th~ low resolution of the peaks is caused by a large amount of noise Furthermore randomly broken chromosomes and nuclear fragments form a broad underlying background continuum [4J which may be increased considerably folshylowing a treatment with environmental mutagens prior to chromosome isolation [17] Therefore it is desirable to reshyduce the noise fluctuations and to eliminate (or to quantify) the amount of background present in the histogram by computer analysis of the stored multichannel data

119

242

Fig3 Data evaluation Smoothing and background elimination ordinate Number of particles abscissa Channel number (corresponding to the logarithm of the DAPI fluorescence intensity) a) Experimental distribution prior to data processing Chromosomes were isolated from the Chinese hamster cell line CHL according to [14 15] and transfered to low ionic strength buffer appropriate for free flow electrophoresis b) Data of Fig 3a following smoothing (noise elishymination) and background subtraction see text for details (from

[ 48J)

For this purpose a variety of mathematical procedures are available With smoothing routines the measured signal may be transformed into a signal with a better signalapshyparent noise ratio According to our experience the so called k-nearest-neighbours method with an appropriately sized

I

330 C Cremer et al Laser in utnrnptlv ApplIcatHms in Flow

interval of abscissa values (2k + 1) seemed to fulfil this transshyformation quite wen for flow karyotypes Here it is assumed that the measured curve cohsists of two parts a true smooth curve and the noise values [18] Within the chosen interval the latter are supposed nearly to sum up to zero Then each ordinate value of the new transformed smooth flow histogram is computed by a weightened sum of the k nearest neighbour ordinate values of the original curve The used weighting factors have the form of binominal factors

aj = (2k + l)jj(k + 1- j) (withj = -k +k)

Under the conditions chosen the area of the total flow karshyyotype changes only less than 01 compared to the original measured histogram

In Fig 3b the result of such a smoothing protocol is shown (applied to the data of Fig 3a) Reducing the apparshyent noise a number of true chromosome peaks is highshylighted

In addition the background was quantified and subshytracted In this case the assumption of a ChP-distribution function resulted in a better background subtraction than the exponential assumption often used (J Doumllle et aL manshyuscript in preparation)

Flow histograms thus obtained may be analyzed further According to theory an ideal linear flow karyotype (without noise and background) is composed of Gaussian distribushytions one for the fluorescence values of each different chroshymosome type [4] The areas of these Gaussians correspond to the relative frequencies of the respective chromosome types in the suspension Peak means and coefficients of varshyiation (CV = standard deviation of the peak divided by peak mean) give precise quantitative information on the composition of the flow karyotype To obtain these data ie composition of the single peaks relative and absolute peak area CV etc flow karyotypes may be approximated as a sum of Gaussian distributions

Considering the rise of the logarithmic amplification used in most flow cytometers it is according to our experience also possible to use the same fit procedure for flow karyoshytypes with a logarithmic distribution along the fluorescence axis In this case the area of a peak typical for a chromoshysome in a logarithmic flow karyotype usually differs less than 1 from the area of the theoretical peak resulting from a Gaussian fit Fig 4 gives an example for this kind of evalshyuation In Fig 4a a flow karyotype of the Chinese hamster cellline CHL is presented after application of the smoothing protocol and background subtraction as described above In Fig 4b a theoretical flow karyotype is shown composed of 10 single Gaussian curves by an interactive PC program Peaks consisting of several chromosome types are separated into single overlapping peaks The resulting curve summing up those peaks fits the measured flow karyotype quite weIl (difIerence in peak area less than 2) Data evaluation procedures such as those described

above may contribute considerably to improve the analysis gt~~~()f~xperimental flow karyotypes However the decompositshyllon into Gaussian distributions requires low background

middotijtJ~ve1s and sufiiciently separated peaks to be reliable fiCli7

Fig4 Matching of Gaussians to a flow karyotype

ordinate Number of partic1es abscissa Channel number (correshysponding to the logarithm of DAPI fluorescence intensity) a) flow karyotype (Chinese hamster cell line CHL) after smoothing and background subtraction (see Fig 3) of the experimental data b) decomposition of the flow karyotype (Fig 4a) into 10 Gaussian fit

curves

Since noise and background levels critically depend on the experimental conditions (eg chromosome isolation proshycedures number of particles measured laser light intensity and stability stability of the fluid flow system electronical set up) a careful adjustment of these parameters may be critical for sufficiently weIl resolved flow karyotypes

Under the conditions described so far the peak separashytion however may not be improved beyond a certain limit ie if two chromosome types have the same total DNA content they cannot be separated by univariate flow cytoshymetry alone using DNA stains which do not prefer either AT or GC base pairs (for instance intercalators like Proshypidium Iodide or Ethidium Bromide) Therefore a signifishycant progress in this context was achieved by using AT GC specific DNA stains (eg Hoechst 33258 Hoechst 33342 Chromomycin A3 Mithramycin etc) together with dual lashyser flow cytometry

Dual Laser Flow Cytometry

To perform dual beam (laser) flow cytometry [19] the isolated chromosomes are stained with two fluorescent dyes instead of one only These two dyes have different base speshy

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 331

cifity and different excitation and emission wavelengths An intensively used dye combination is Hoechst 33258 (HO) wruch is AT specific and Chromomycin A3 (CA3) which is GC specific

In a dual laser flow cytometer the stained chromosomes pass one after another two slightly elliptically focused CW laser beams of different wavelengths In case of the fluoroshychromes named above UV multiline excitation is used for HO blue laser light (mainly 458 nm) is used to excite CA3 fluorescence The fluorescence emissions are collected inshydependently from each other This is possible due to the spatial distance of the two laser intersection points Thus information may be obtained not only on the relative DNA content (06 x HO + 04 x CA3 fluorescence intensity) [26J but on the AT (HO fluorescence)GC (CA3 fluoresshycence) ratio as weIl

4)

Fig5 Bivariate distribution (dual laser flow cytometry) of chromosomes isolated from peripheral human lymphocytes After removal of the blood the chromosomes were isolated following 90 min treatment with demecolcine (Colcemid) and then measured with the Livershymore dual laser flow cytometer [19] No noise level or background elimination procedure was applied ordinate HO fluorescence inshytensity (64 channels UV multiline excitation) abscissa CA3 fluoshyrescence intensity (64 channels 458 nm excitation) The fluorescence amplification was linear for both parameters The contour lines of equal particle frequencies are shown (relative levels 1006228 13 6 3 1 05) The peak identification was made according to [20] Due to the fluorescence gain used the large chromosomes No 1 23 were excluded from the display This allowed a better discrimshy

ination of the smaller chromosomes

Fig5 shows an application of dual laser flow cytometry in the quantitative analysis of human chromosomes Here chromosomes were isolated [20J from a short term culture (72 hours) of female peripheral blood (normal donor) The ordinate of this bivariate plot shows the fluorescence of the AT specific dye HO the abscissa presents the fluoresshycence of the GC specific dye CA3 The contour curves inshydicate the lines of equal partic1e frequency Therefore c10sed contour lines represent a weIl separated chromosome peak All of the 20 different chromosome types measured he re (No 4-22 X) may be assigned [20] In this example eleven of

them (chromosome types No 4 5 6 13 16 17 18 19 20- 21 22) are represented by individual chromosome pe~ ~~ Using dual laser flow cytometrymiddot togetherwith the fluor~i~ chromes HO and CA3 it is possible to discriminate all hu-i~ man chromosomes [20] except chromosomes No 9-12 which are too similar both in total DNA content and in ATGC ratio Some further progress in the resolution of the 9 -12 peak is possible using the fluorescence dye DIPI (preshyfering AT base pairs [21]) instead of HO

Using the sorting procedure described above individual human chromosome types have been purified by dual beam sorting with efficiencies up to 97 [21] Instead of sorting chromosomes isolated from human cells it is also possible to sort human chromosomes isolated from Chinese hamster x human hybrid cells which contain a reduced number of human chromosome types [122223] In a number of cases this allows an increased purification

Improvement of Flow Cytometric Resolution of Specific Chromosome Types by Chemical Modification of DNA

Under favourable circumstances it is nowadays possible to detect quantitatively changes in DNA content as smaIl as 10- 15 g [4]

However there are problems which may requiie addishytional discrimination For example it may be interesting to analyze the chromosomes of other mammals or to improve the flow cytometric analysis of chromosomes isolated from tumor cells which often show a variety of complex rearshyrangements

A) Incorporation of Base Analogues

One possibility to obtain an additional discrimination of specific chromosome types [24 25J is to use the fact that the chromosomal DNA may be replicated at different times of the cell cycle

The Chinese hamster Y chromosome for example is the last to be replicated If the base analogue 5-bromo-2-deoxyshyuridine (BrdUrd) is incorporated instead of thymidine durshying the last hours prior to mitosis then the AT -specific flushyorescence of the dye Hoechst 33258 is quenched I

Fig 6 gives an example [25] Here bivariate contour plots are shown as indicated in Fig5 The only difference is that in this case chromosomes of a Chinese hamster cellline (M3-

11) [10 26] were used instead of human chromosomes Fig 6a presents the dual parameter flow karyotype of these IChinese hamster chromosomes after HOCA3 staining but without any other modification Here the peak of the Y I chromosomes is not weH separated from neighbouring chroshymosome peaks A microscopic analysis of particles sorted from this Y position revealed that only 15 of the sorted chromosomes were identified as Y chromosomes [24J Such I low purities are insufficient to establish for example a chroshymosome specific DNA library [12 23 27J or to perform other studies eg gene assignment [21] In the two contour plots shown in Fig 6b c BrdUrd was added to the culture I medium 25 hrs (Fig 6b) and 35 hrs (Fig 6c) prior to chroshy I mosome isolation (90 min Colcemid treatment for collection I

of mitotic cells) In this case the HO fluorescence of the Y I chromosome (Ya) was increasingly quenched while the CA3 fluorescence remained about the same Due to this it became

332 C Cremer et al Laser in t t middot Applicati~)ns in Flow

11 OhnBrdU

~ bull xS)

67

(bI 2S hn BrdU

bull fmiddot MI I

Va

10M27 (c) 3S hrsBrdU

Mt ~bullbullxS) 5

9shy

4S8 nm eaCltdtlon (CAJ lIuorlllcnce)

Fig6 Effect of a terminal label with BrdUrd on the bivariate flow karshyyotype of HO - CA3 stained chromosomes M 3-1 cells were incushybated for 0 (a) 25 hrs (b) and 35 hrs (c) with BrdUrd before isolation Chromosome assignment was done according to [26J except for the Y which was identified by dual parameter sorting on microscope slides followed by Q-banding analysis Experimental settings were chosen to highlight the chromosome regions of intershyest (Y 10 11 M2) This resulted in chromosomes 1 and 2 being eliminated from the display Experimental conditions were held

constant for all measurements (from [25J)

possible to c1early separate the position of the Ya peakfrom all other chromosomes In case of the condition shown in

fig 6c~the Y chromosome was sorted with relatively high 1gtf~i~ J7 ) ~~lt ~ ~ ~) ~ ~ ~

~~)J~~fY~0bri~etri~ ~etection ~f Chromosomes Following situ Hy degdizatioDin Suspeosion

- ~lC~~ Ii~ gt - gt lt

dB~Urdapproachdiscussed above is one possibility ~pccmc physico-chemica1 DNA modification may be

Iff4A uOcd to improve the middotszligow cytometric discrimination of

certain chromosome types In practice the BrdUrd method appears to be useful for chromosome types with grossly different replication kinetics [25]

In this section a new approach will be described which in principle may be applied to any chromosome type This new technique is based on the modification of the DNA of specific chromosomes by in situ hybridization with specific DNA probes These probes have been chemically modified eg biotinylated by nick translation then they are hybridshyized to the chromosomes This technique requires denaturshyation (usually done by thermal treatment) of the chromoshysomal DNA prior to hybridization (denaturation temperashyture is typically about 73degC) ie the two native DNA strands are physically separated from each other After renaturation a new double stranded DNA is formed with the complementary sequence of the DNA probe

The biotinylated DNA strands hybridized to the chroshymosomal DNA sequences can be detected eg by a specific fluorcscing antibody detection system (fluorescence hybridshyization chromosome painting) or an enzyme detection reaction

While this technique has already being sucessfully develshyoped for the analysis of chromosomes fixed on slides [28 - 35] its application to flow cytometry of chromosomes is still in the very beginning

Since in this case a suspension of isolated chromosomes is required thc entire fluorescence hybridization protocol has to be performed in suspension maintaining the morshyphological stability of the chromosomes A critical step of this method is the denaturation and renaturation of the chromosomes in suspension This process may be monitored by flow cytometry In the results presented in Fig 7 flow karyotypes of the chromosomes of a Chinese hamster cell line are shown shortly before thermal denaturation (Fig 7a) and at different times afterwards (Figs 7b c d) Immediately following denaturation the temperature was lowered(room temperature) so that reannealing of the two strands was possible For flow cytometry the chromosomes were stained with DAPI a fluorescent dye preferably binding to double stranded (Le renatured) DNA

All flow cytometric measurements prescnted in Fig 7 were performed under the identical instrumental set up (fluoresshycence gain etc) aliquots from the same DAPI stained chroshymosome suspension were used Thc chromosomes of the flow karyotypes in Fig 7b c d were additionally stabilized with propidium iodide (PI) This led to q uenching of the DAPI fluorescence (quantum yield about 50) Thus the flow distributions obtained may be compared with each other

Two minutes after denaturation (Fig 7b) the flow distrishybution was very poorly resolved The DAPI fluorescence of aB chromosomes was found to be significantly reduced This suggests that under the conditions used a considerable amount of denaturation took place With increasing time however the original flow karyotype (Fig 7a) gradually reappeared (Figs 7c d)

lt is interesting to note that in the course of the reanshynealing process the fluorescence intensities (the effect of PI taken into consideration) of the individual chromosome peaks regain values alm ost identical to those prior to deshy

-

C Cremer ct aL Laser in trllTlptr AP1PllcaUC)ns in Flow328

search laboratories and medical institutions For example flow cytometric measuretnents of cellular DNA content alshylow the quantitative characterization of a variety of tumor cell lines and thus have found a vast application in tumor diagnosis [2 3J

This report will focus on some aspects of the use of flow cytometry for quantitative characterization of chromosomes rltflow cytogenetics) Since a general review on this field has been published recently [4] this artic1e in particular will discuss in more detail some approaches to increase the resshyolution of flow cytometric data

Principles of Laser Fluorescence Acthated Analysis and Sorting

The technique of laser fluorescence activated analysis and sorting of cells and chromosomes has mainly been developed in the last two decades [5J and has become important for many applications both in biomedical research and c1inical diagnosis (for general reviews see [6 - 9J)

Schematic design of a single laser beam flow cytometer and sorter See text for details (by courtesy of Coulter Corp)

Its least complex form is presented in Fig 1 A suspension of partic1es to be measured for instance individual cells or isolated chromosomes is injected coaxially into the sheath fluid stream of the flow chamber (flow celI) By means of hydrodynamic focusing through a jewel orifice of 50- 200 Jlm diameter (typical diameter for chromosome analysis 70- 80 JLm) at the exit of the flow cell the particles are confined within the core of the stream of only a few micrometers diameter (typically about 10 of the jet diashy

passing one by one a slightly focused CW laser beam Tbe high flux excites the particles stained with fluoroshychromes specific for example to DNA DNA base ratio or cell surface antigens

In most cases argon- and krypton-ion lasers (excitation of dyes in the visible and near UV) are used with total emission power of several watts For some immunological applications helium-neon or ion-pumped dye lasers are used too A high stability is required for precise measureshyments because for small partic1es as chromosomes any varshyiation in laser power may result in a variation of the emitted fluorescence intensity

For eaeh partic1e the fluoreseence emission is deteeted by photomultipliers (PMT) during the few microseeonds taken to traverse the foeal volume The PMT signal is either linshyeally or logarithmieally integrated digitized and registered by a multi-ehannel analyzer The larger for instance the DNA content of a cell nuc1eus or a chromosome is the more dye may be bound and consequently the fluorescence intenshysity inereases Thus it is possible to discriminate different types of partic1es aceording to their fluorescence properties

Using a single laser be am device two fluoreseenee parashymeters (following simultaneous staining with two fluoroshychromes of similar absorption spectra) may be measured if the fluorescence emission speetra of the two dyes do not overlap In this ca se e g green and red fluoreseenee are separated by appropriate edge or band pass filters In adshydition the laser light scattered by a partic1e mayaiso be measured and used for further diserimination For this forshyward and right angle light seatter is detectable Nowadays many commercial machines are equipped with more than four detectors and two lasers for two three or multicolor fluorescenee measurements

For physical separation of specific types of particles (sorshyting) a hydrodynamic electronie procedure is applied which allows to sort them into collection vessels aecording to preshyseleeted optieal properties measured as deseribed above To do this the fluid jet in air is disrupted into single droplets (diameter about 18 of the jet diameter) after having passed the laser intersection point This is realized by mechanical vibrations of the flow cell up to 40 kHz in vertical direction parallel to the streaming jet Those droplets eontaining parshyticles to be sorted are charged electrically before separation from the jet and deflected into recipient vials by an eleetroshystatic field of a few kVjem As mentioned above the method presently allows analysis and sorting of particles between several hundreds and 20000 per second

Flow Cytometry of Chromosomes Univariate Distributions

In this report we shall foeus on a special feature of laser fluorescenee activated analysis and sorting i e flow cytoshymetry of isolated metaphase chromosomes in suspension (also called flow cytogeneties) ([10J for review see [4J) This has found important applications in moleeular biology human genetics and c1inical diagnostics of genetic diseases Presently one of the main applieations of chromosome sortshying is the construction of chromosome specific DNA librarshyies [11 12J

Fig 2 gives an exampIe of a so called univariate flow karyotype using one fluoreseenee parameter only On the ordinate the number of the fluorescence signals (partic1es) is given on the abscissa the logarithm of the fluorescence inshytensity is plotted Ideally eaeh fluorescence signal represents

I

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 329

Imiddot a

(

3M I

200

192

96

8 -1 161 242

Fig2 U nivariate flow karyotype of chromosomes from the Chinese hamshyster cell line CHL (DAPI staining laser excitation in the near UV) ordinate Number of particles abscissa Channel number (in this case corresponding to the logarithm of the fluorescence intensity) The peak designation (No 1 - 9) was made according to decreasing fluorescence intensity (correlated with the chromosome length in case of chromosomes of the Chinese hamster) The flow karyotype shown has been obtained without further processing of the data

one chromosome with a specific DNA conte nt In the case shown in Fig 2 chromosomes isolated from a Chinese hamshyster cell line were used The chromosome suspension was stained with the fluorochrome DAPI (5 JlM) specific for doublestranded DNA Excitation was done by an argon-ion laser (UV multiline 3511 - 3638 nm) with 50 mW Under favourable circumstances and optimum resolution each chromosome type of a cell line is represented by one indishyvidual peak One possibility to assign the peaks to individual chromosome types is to sort the peak contents and to idenshytify the chromosomes by conventional cytogenetic methods [1013] Another preliminary assignment is usually possible by comparison of the flow cytogenetic data with light mishycroscopic observations Under appropriate conditions ie that the ATGC correlation is about the same for each chroshymosome type an excellent correlation may be obtained beshytween chromosome length and relative fluorescence intensity of the flow karyotype

To achieve the flow karyotype shown in Fig2 a chroshymosome isolation method [14 15J was applied using a hexshyandiol isolation buffer to stabilize the chromosomes This procedure has the particular advantage that chromosomes in suspension may be stored at 4degC for long times (in our experience up to 5 years) without a significant deterioration of the resolution of the flow karyotype [16]

Data Evaluation

In many cases the flow karyotypes obtained under various experimental circumstances are not as weIl resolved as shown in Fig 2 This may be caused by different preparation conditions one of them is the buffer system of the chroshymosome suspension which may not always be optimal for flow karyotyping but necessary with respect to other expershyimental studies done with the chromosomes

For example free flow electrophoresis of isolated metashyphase chromosomes requires a buffer system of low ionic

strength [48] Flow cytometric characterization mayJeadi~~ to poorIy resolved flow karyotypes because of morpbolo~~r~ ical instabilities of the chromosomes in this buffer ili~fl

In Fig 3a the experimental flow karyotype of chromc)~i ~ somes of a Chinese hamster cell line in free flow electrb~ phoresis buffer prior to data processing is presented~Th~ low resolution of the peaks is caused by a large amount of noise Furthermore randomly broken chromosomes and nuclear fragments form a broad underlying background continuum [4J which may be increased considerably folshylowing a treatment with environmental mutagens prior to chromosome isolation [17] Therefore it is desirable to reshyduce the noise fluctuations and to eliminate (or to quantify) the amount of background present in the histogram by computer analysis of the stored multichannel data

119

242

Fig3 Data evaluation Smoothing and background elimination ordinate Number of particles abscissa Channel number (corresponding to the logarithm of the DAPI fluorescence intensity) a) Experimental distribution prior to data processing Chromosomes were isolated from the Chinese hamster cell line CHL according to [14 15] and transfered to low ionic strength buffer appropriate for free flow electrophoresis b) Data of Fig 3a following smoothing (noise elishymination) and background subtraction see text for details (from

[ 48J)

For this purpose a variety of mathematical procedures are available With smoothing routines the measured signal may be transformed into a signal with a better signalapshyparent noise ratio According to our experience the so called k-nearest-neighbours method with an appropriately sized

I

330 C Cremer et al Laser in utnrnptlv ApplIcatHms in Flow

interval of abscissa values (2k + 1) seemed to fulfil this transshyformation quite wen for flow karyotypes Here it is assumed that the measured curve cohsists of two parts a true smooth curve and the noise values [18] Within the chosen interval the latter are supposed nearly to sum up to zero Then each ordinate value of the new transformed smooth flow histogram is computed by a weightened sum of the k nearest neighbour ordinate values of the original curve The used weighting factors have the form of binominal factors

aj = (2k + l)jj(k + 1- j) (withj = -k +k)

Under the conditions chosen the area of the total flow karshyyotype changes only less than 01 compared to the original measured histogram

In Fig 3b the result of such a smoothing protocol is shown (applied to the data of Fig 3a) Reducing the apparshyent noise a number of true chromosome peaks is highshylighted

In addition the background was quantified and subshytracted In this case the assumption of a ChP-distribution function resulted in a better background subtraction than the exponential assumption often used (J Doumllle et aL manshyuscript in preparation)

Flow histograms thus obtained may be analyzed further According to theory an ideal linear flow karyotype (without noise and background) is composed of Gaussian distribushytions one for the fluorescence values of each different chroshymosome type [4] The areas of these Gaussians correspond to the relative frequencies of the respective chromosome types in the suspension Peak means and coefficients of varshyiation (CV = standard deviation of the peak divided by peak mean) give precise quantitative information on the composition of the flow karyotype To obtain these data ie composition of the single peaks relative and absolute peak area CV etc flow karyotypes may be approximated as a sum of Gaussian distributions

Considering the rise of the logarithmic amplification used in most flow cytometers it is according to our experience also possible to use the same fit procedure for flow karyoshytypes with a logarithmic distribution along the fluorescence axis In this case the area of a peak typical for a chromoshysome in a logarithmic flow karyotype usually differs less than 1 from the area of the theoretical peak resulting from a Gaussian fit Fig 4 gives an example for this kind of evalshyuation In Fig 4a a flow karyotype of the Chinese hamster cellline CHL is presented after application of the smoothing protocol and background subtraction as described above In Fig 4b a theoretical flow karyotype is shown composed of 10 single Gaussian curves by an interactive PC program Peaks consisting of several chromosome types are separated into single overlapping peaks The resulting curve summing up those peaks fits the measured flow karyotype quite weIl (difIerence in peak area less than 2) Data evaluation procedures such as those described

above may contribute considerably to improve the analysis gt~~~()f~xperimental flow karyotypes However the decompositshyllon into Gaussian distributions requires low background

middotijtJ~ve1s and sufiiciently separated peaks to be reliable fiCli7

Fig4 Matching of Gaussians to a flow karyotype

ordinate Number of partic1es abscissa Channel number (correshysponding to the logarithm of DAPI fluorescence intensity) a) flow karyotype (Chinese hamster cell line CHL) after smoothing and background subtraction (see Fig 3) of the experimental data b) decomposition of the flow karyotype (Fig 4a) into 10 Gaussian fit

curves

Since noise and background levels critically depend on the experimental conditions (eg chromosome isolation proshycedures number of particles measured laser light intensity and stability stability of the fluid flow system electronical set up) a careful adjustment of these parameters may be critical for sufficiently weIl resolved flow karyotypes

Under the conditions described so far the peak separashytion however may not be improved beyond a certain limit ie if two chromosome types have the same total DNA content they cannot be separated by univariate flow cytoshymetry alone using DNA stains which do not prefer either AT or GC base pairs (for instance intercalators like Proshypidium Iodide or Ethidium Bromide) Therefore a signifishycant progress in this context was achieved by using AT GC specific DNA stains (eg Hoechst 33258 Hoechst 33342 Chromomycin A3 Mithramycin etc) together with dual lashyser flow cytometry

Dual Laser Flow Cytometry

To perform dual beam (laser) flow cytometry [19] the isolated chromosomes are stained with two fluorescent dyes instead of one only These two dyes have different base speshy

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 331

cifity and different excitation and emission wavelengths An intensively used dye combination is Hoechst 33258 (HO) wruch is AT specific and Chromomycin A3 (CA3) which is GC specific

In a dual laser flow cytometer the stained chromosomes pass one after another two slightly elliptically focused CW laser beams of different wavelengths In case of the fluoroshychromes named above UV multiline excitation is used for HO blue laser light (mainly 458 nm) is used to excite CA3 fluorescence The fluorescence emissions are collected inshydependently from each other This is possible due to the spatial distance of the two laser intersection points Thus information may be obtained not only on the relative DNA content (06 x HO + 04 x CA3 fluorescence intensity) [26J but on the AT (HO fluorescence)GC (CA3 fluoresshycence) ratio as weIl

4)

Fig5 Bivariate distribution (dual laser flow cytometry) of chromosomes isolated from peripheral human lymphocytes After removal of the blood the chromosomes were isolated following 90 min treatment with demecolcine (Colcemid) and then measured with the Livershymore dual laser flow cytometer [19] No noise level or background elimination procedure was applied ordinate HO fluorescence inshytensity (64 channels UV multiline excitation) abscissa CA3 fluoshyrescence intensity (64 channels 458 nm excitation) The fluorescence amplification was linear for both parameters The contour lines of equal particle frequencies are shown (relative levels 1006228 13 6 3 1 05) The peak identification was made according to [20] Due to the fluorescence gain used the large chromosomes No 1 23 were excluded from the display This allowed a better discrimshy

ination of the smaller chromosomes

Fig5 shows an application of dual laser flow cytometry in the quantitative analysis of human chromosomes Here chromosomes were isolated [20J from a short term culture (72 hours) of female peripheral blood (normal donor) The ordinate of this bivariate plot shows the fluorescence of the AT specific dye HO the abscissa presents the fluoresshycence of the GC specific dye CA3 The contour curves inshydicate the lines of equal partic1e frequency Therefore c10sed contour lines represent a weIl separated chromosome peak All of the 20 different chromosome types measured he re (No 4-22 X) may be assigned [20] In this example eleven of

them (chromosome types No 4 5 6 13 16 17 18 19 20- 21 22) are represented by individual chromosome pe~ ~~ Using dual laser flow cytometrymiddot togetherwith the fluor~i~ chromes HO and CA3 it is possible to discriminate all hu-i~ man chromosomes [20] except chromosomes No 9-12 which are too similar both in total DNA content and in ATGC ratio Some further progress in the resolution of the 9 -12 peak is possible using the fluorescence dye DIPI (preshyfering AT base pairs [21]) instead of HO

Using the sorting procedure described above individual human chromosome types have been purified by dual beam sorting with efficiencies up to 97 [21] Instead of sorting chromosomes isolated from human cells it is also possible to sort human chromosomes isolated from Chinese hamster x human hybrid cells which contain a reduced number of human chromosome types [122223] In a number of cases this allows an increased purification

Improvement of Flow Cytometric Resolution of Specific Chromosome Types by Chemical Modification of DNA

Under favourable circumstances it is nowadays possible to detect quantitatively changes in DNA content as smaIl as 10- 15 g [4]

However there are problems which may requiie addishytional discrimination For example it may be interesting to analyze the chromosomes of other mammals or to improve the flow cytometric analysis of chromosomes isolated from tumor cells which often show a variety of complex rearshyrangements

A) Incorporation of Base Analogues

One possibility to obtain an additional discrimination of specific chromosome types [24 25J is to use the fact that the chromosomal DNA may be replicated at different times of the cell cycle

The Chinese hamster Y chromosome for example is the last to be replicated If the base analogue 5-bromo-2-deoxyshyuridine (BrdUrd) is incorporated instead of thymidine durshying the last hours prior to mitosis then the AT -specific flushyorescence of the dye Hoechst 33258 is quenched I

Fig 6 gives an example [25] Here bivariate contour plots are shown as indicated in Fig5 The only difference is that in this case chromosomes of a Chinese hamster cellline (M3-

11) [10 26] were used instead of human chromosomes Fig 6a presents the dual parameter flow karyotype of these IChinese hamster chromosomes after HOCA3 staining but without any other modification Here the peak of the Y I chromosomes is not weH separated from neighbouring chroshymosome peaks A microscopic analysis of particles sorted from this Y position revealed that only 15 of the sorted chromosomes were identified as Y chromosomes [24J Such I low purities are insufficient to establish for example a chroshymosome specific DNA library [12 23 27J or to perform other studies eg gene assignment [21] In the two contour plots shown in Fig 6b c BrdUrd was added to the culture I medium 25 hrs (Fig 6b) and 35 hrs (Fig 6c) prior to chroshy I mosome isolation (90 min Colcemid treatment for collection I

of mitotic cells) In this case the HO fluorescence of the Y I chromosome (Ya) was increasingly quenched while the CA3 fluorescence remained about the same Due to this it became

332 C Cremer et al Laser in t t middot Applicati~)ns in Flow

11 OhnBrdU

~ bull xS)

67

(bI 2S hn BrdU

bull fmiddot MI I

Va

10M27 (c) 3S hrsBrdU

Mt ~bullbullxS) 5

9shy

4S8 nm eaCltdtlon (CAJ lIuorlllcnce)

Fig6 Effect of a terminal label with BrdUrd on the bivariate flow karshyyotype of HO - CA3 stained chromosomes M 3-1 cells were incushybated for 0 (a) 25 hrs (b) and 35 hrs (c) with BrdUrd before isolation Chromosome assignment was done according to [26J except for the Y which was identified by dual parameter sorting on microscope slides followed by Q-banding analysis Experimental settings were chosen to highlight the chromosome regions of intershyest (Y 10 11 M2) This resulted in chromosomes 1 and 2 being eliminated from the display Experimental conditions were held

constant for all measurements (from [25J)

possible to c1early separate the position of the Ya peakfrom all other chromosomes In case of the condition shown in

fig 6c~the Y chromosome was sorted with relatively high 1gtf~i~ J7 ) ~~lt ~ ~ ~) ~ ~ ~

~~)J~~fY~0bri~etri~ ~etection ~f Chromosomes Following situ Hy degdizatioDin Suspeosion

- ~lC~~ Ii~ gt - gt lt

dB~Urdapproachdiscussed above is one possibility ~pccmc physico-chemica1 DNA modification may be

Iff4A uOcd to improve the middotszligow cytometric discrimination of

certain chromosome types In practice the BrdUrd method appears to be useful for chromosome types with grossly different replication kinetics [25]

In this section a new approach will be described which in principle may be applied to any chromosome type This new technique is based on the modification of the DNA of specific chromosomes by in situ hybridization with specific DNA probes These probes have been chemically modified eg biotinylated by nick translation then they are hybridshyized to the chromosomes This technique requires denaturshyation (usually done by thermal treatment) of the chromoshysomal DNA prior to hybridization (denaturation temperashyture is typically about 73degC) ie the two native DNA strands are physically separated from each other After renaturation a new double stranded DNA is formed with the complementary sequence of the DNA probe

The biotinylated DNA strands hybridized to the chroshymosomal DNA sequences can be detected eg by a specific fluorcscing antibody detection system (fluorescence hybridshyization chromosome painting) or an enzyme detection reaction

While this technique has already being sucessfully develshyoped for the analysis of chromosomes fixed on slides [28 - 35] its application to flow cytometry of chromosomes is still in the very beginning

Since in this case a suspension of isolated chromosomes is required thc entire fluorescence hybridization protocol has to be performed in suspension maintaining the morshyphological stability of the chromosomes A critical step of this method is the denaturation and renaturation of the chromosomes in suspension This process may be monitored by flow cytometry In the results presented in Fig 7 flow karyotypes of the chromosomes of a Chinese hamster cell line are shown shortly before thermal denaturation (Fig 7a) and at different times afterwards (Figs 7b c d) Immediately following denaturation the temperature was lowered(room temperature) so that reannealing of the two strands was possible For flow cytometry the chromosomes were stained with DAPI a fluorescent dye preferably binding to double stranded (Le renatured) DNA

All flow cytometric measurements prescnted in Fig 7 were performed under the identical instrumental set up (fluoresshycence gain etc) aliquots from the same DAPI stained chroshymosome suspension were used Thc chromosomes of the flow karyotypes in Fig 7b c d were additionally stabilized with propidium iodide (PI) This led to q uenching of the DAPI fluorescence (quantum yield about 50) Thus the flow distributions obtained may be compared with each other

Two minutes after denaturation (Fig 7b) the flow distrishybution was very poorly resolved The DAPI fluorescence of aB chromosomes was found to be significantly reduced This suggests that under the conditions used a considerable amount of denaturation took place With increasing time however the original flow karyotype (Fig 7a) gradually reappeared (Figs 7c d)

lt is interesting to note that in the course of the reanshynealing process the fluorescence intensities (the effect of PI taken into consideration) of the individual chromosome peaks regain values alm ost identical to those prior to deshy

-

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 329

Imiddot a

(

3M I

200

192

96

8 -1 161 242

Fig2 U nivariate flow karyotype of chromosomes from the Chinese hamshyster cell line CHL (DAPI staining laser excitation in the near UV) ordinate Number of particles abscissa Channel number (in this case corresponding to the logarithm of the fluorescence intensity) The peak designation (No 1 - 9) was made according to decreasing fluorescence intensity (correlated with the chromosome length in case of chromosomes of the Chinese hamster) The flow karyotype shown has been obtained without further processing of the data

one chromosome with a specific DNA conte nt In the case shown in Fig 2 chromosomes isolated from a Chinese hamshyster cell line were used The chromosome suspension was stained with the fluorochrome DAPI (5 JlM) specific for doublestranded DNA Excitation was done by an argon-ion laser (UV multiline 3511 - 3638 nm) with 50 mW Under favourable circumstances and optimum resolution each chromosome type of a cell line is represented by one indishyvidual peak One possibility to assign the peaks to individual chromosome types is to sort the peak contents and to idenshytify the chromosomes by conventional cytogenetic methods [1013] Another preliminary assignment is usually possible by comparison of the flow cytogenetic data with light mishycroscopic observations Under appropriate conditions ie that the ATGC correlation is about the same for each chroshymosome type an excellent correlation may be obtained beshytween chromosome length and relative fluorescence intensity of the flow karyotype

To achieve the flow karyotype shown in Fig2 a chroshymosome isolation method [14 15J was applied using a hexshyandiol isolation buffer to stabilize the chromosomes This procedure has the particular advantage that chromosomes in suspension may be stored at 4degC for long times (in our experience up to 5 years) without a significant deterioration of the resolution of the flow karyotype [16]

Data Evaluation

In many cases the flow karyotypes obtained under various experimental circumstances are not as weIl resolved as shown in Fig 2 This may be caused by different preparation conditions one of them is the buffer system of the chroshymosome suspension which may not always be optimal for flow karyotyping but necessary with respect to other expershyimental studies done with the chromosomes

For example free flow electrophoresis of isolated metashyphase chromosomes requires a buffer system of low ionic

strength [48] Flow cytometric characterization mayJeadi~~ to poorIy resolved flow karyotypes because of morpbolo~~r~ ical instabilities of the chromosomes in this buffer ili~fl

In Fig 3a the experimental flow karyotype of chromc)~i ~ somes of a Chinese hamster cell line in free flow electrb~ phoresis buffer prior to data processing is presented~Th~ low resolution of the peaks is caused by a large amount of noise Furthermore randomly broken chromosomes and nuclear fragments form a broad underlying background continuum [4J which may be increased considerably folshylowing a treatment with environmental mutagens prior to chromosome isolation [17] Therefore it is desirable to reshyduce the noise fluctuations and to eliminate (or to quantify) the amount of background present in the histogram by computer analysis of the stored multichannel data

119

242

Fig3 Data evaluation Smoothing and background elimination ordinate Number of particles abscissa Channel number (corresponding to the logarithm of the DAPI fluorescence intensity) a) Experimental distribution prior to data processing Chromosomes were isolated from the Chinese hamster cell line CHL according to [14 15] and transfered to low ionic strength buffer appropriate for free flow electrophoresis b) Data of Fig 3a following smoothing (noise elishymination) and background subtraction see text for details (from

[ 48J)

For this purpose a variety of mathematical procedures are available With smoothing routines the measured signal may be transformed into a signal with a better signalapshyparent noise ratio According to our experience the so called k-nearest-neighbours method with an appropriately sized

I

330 C Cremer et al Laser in utnrnptlv ApplIcatHms in Flow

interval of abscissa values (2k + 1) seemed to fulfil this transshyformation quite wen for flow karyotypes Here it is assumed that the measured curve cohsists of two parts a true smooth curve and the noise values [18] Within the chosen interval the latter are supposed nearly to sum up to zero Then each ordinate value of the new transformed smooth flow histogram is computed by a weightened sum of the k nearest neighbour ordinate values of the original curve The used weighting factors have the form of binominal factors

aj = (2k + l)jj(k + 1- j) (withj = -k +k)

Under the conditions chosen the area of the total flow karshyyotype changes only less than 01 compared to the original measured histogram

In Fig 3b the result of such a smoothing protocol is shown (applied to the data of Fig 3a) Reducing the apparshyent noise a number of true chromosome peaks is highshylighted

In addition the background was quantified and subshytracted In this case the assumption of a ChP-distribution function resulted in a better background subtraction than the exponential assumption often used (J Doumllle et aL manshyuscript in preparation)

Flow histograms thus obtained may be analyzed further According to theory an ideal linear flow karyotype (without noise and background) is composed of Gaussian distribushytions one for the fluorescence values of each different chroshymosome type [4] The areas of these Gaussians correspond to the relative frequencies of the respective chromosome types in the suspension Peak means and coefficients of varshyiation (CV = standard deviation of the peak divided by peak mean) give precise quantitative information on the composition of the flow karyotype To obtain these data ie composition of the single peaks relative and absolute peak area CV etc flow karyotypes may be approximated as a sum of Gaussian distributions

Considering the rise of the logarithmic amplification used in most flow cytometers it is according to our experience also possible to use the same fit procedure for flow karyoshytypes with a logarithmic distribution along the fluorescence axis In this case the area of a peak typical for a chromoshysome in a logarithmic flow karyotype usually differs less than 1 from the area of the theoretical peak resulting from a Gaussian fit Fig 4 gives an example for this kind of evalshyuation In Fig 4a a flow karyotype of the Chinese hamster cellline CHL is presented after application of the smoothing protocol and background subtraction as described above In Fig 4b a theoretical flow karyotype is shown composed of 10 single Gaussian curves by an interactive PC program Peaks consisting of several chromosome types are separated into single overlapping peaks The resulting curve summing up those peaks fits the measured flow karyotype quite weIl (difIerence in peak area less than 2) Data evaluation procedures such as those described

above may contribute considerably to improve the analysis gt~~~()f~xperimental flow karyotypes However the decompositshyllon into Gaussian distributions requires low background

middotijtJ~ve1s and sufiiciently separated peaks to be reliable fiCli7

Fig4 Matching of Gaussians to a flow karyotype

ordinate Number of partic1es abscissa Channel number (correshysponding to the logarithm of DAPI fluorescence intensity) a) flow karyotype (Chinese hamster cell line CHL) after smoothing and background subtraction (see Fig 3) of the experimental data b) decomposition of the flow karyotype (Fig 4a) into 10 Gaussian fit

curves

Since noise and background levels critically depend on the experimental conditions (eg chromosome isolation proshycedures number of particles measured laser light intensity and stability stability of the fluid flow system electronical set up) a careful adjustment of these parameters may be critical for sufficiently weIl resolved flow karyotypes

Under the conditions described so far the peak separashytion however may not be improved beyond a certain limit ie if two chromosome types have the same total DNA content they cannot be separated by univariate flow cytoshymetry alone using DNA stains which do not prefer either AT or GC base pairs (for instance intercalators like Proshypidium Iodide or Ethidium Bromide) Therefore a signifishycant progress in this context was achieved by using AT GC specific DNA stains (eg Hoechst 33258 Hoechst 33342 Chromomycin A3 Mithramycin etc) together with dual lashyser flow cytometry

Dual Laser Flow Cytometry

To perform dual beam (laser) flow cytometry [19] the isolated chromosomes are stained with two fluorescent dyes instead of one only These two dyes have different base speshy

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 331

cifity and different excitation and emission wavelengths An intensively used dye combination is Hoechst 33258 (HO) wruch is AT specific and Chromomycin A3 (CA3) which is GC specific

In a dual laser flow cytometer the stained chromosomes pass one after another two slightly elliptically focused CW laser beams of different wavelengths In case of the fluoroshychromes named above UV multiline excitation is used for HO blue laser light (mainly 458 nm) is used to excite CA3 fluorescence The fluorescence emissions are collected inshydependently from each other This is possible due to the spatial distance of the two laser intersection points Thus information may be obtained not only on the relative DNA content (06 x HO + 04 x CA3 fluorescence intensity) [26J but on the AT (HO fluorescence)GC (CA3 fluoresshycence) ratio as weIl

4)

Fig5 Bivariate distribution (dual laser flow cytometry) of chromosomes isolated from peripheral human lymphocytes After removal of the blood the chromosomes were isolated following 90 min treatment with demecolcine (Colcemid) and then measured with the Livershymore dual laser flow cytometer [19] No noise level or background elimination procedure was applied ordinate HO fluorescence inshytensity (64 channels UV multiline excitation) abscissa CA3 fluoshyrescence intensity (64 channels 458 nm excitation) The fluorescence amplification was linear for both parameters The contour lines of equal particle frequencies are shown (relative levels 1006228 13 6 3 1 05) The peak identification was made according to [20] Due to the fluorescence gain used the large chromosomes No 1 23 were excluded from the display This allowed a better discrimshy

ination of the smaller chromosomes

Fig5 shows an application of dual laser flow cytometry in the quantitative analysis of human chromosomes Here chromosomes were isolated [20J from a short term culture (72 hours) of female peripheral blood (normal donor) The ordinate of this bivariate plot shows the fluorescence of the AT specific dye HO the abscissa presents the fluoresshycence of the GC specific dye CA3 The contour curves inshydicate the lines of equal partic1e frequency Therefore c10sed contour lines represent a weIl separated chromosome peak All of the 20 different chromosome types measured he re (No 4-22 X) may be assigned [20] In this example eleven of

them (chromosome types No 4 5 6 13 16 17 18 19 20- 21 22) are represented by individual chromosome pe~ ~~ Using dual laser flow cytometrymiddot togetherwith the fluor~i~ chromes HO and CA3 it is possible to discriminate all hu-i~ man chromosomes [20] except chromosomes No 9-12 which are too similar both in total DNA content and in ATGC ratio Some further progress in the resolution of the 9 -12 peak is possible using the fluorescence dye DIPI (preshyfering AT base pairs [21]) instead of HO

Using the sorting procedure described above individual human chromosome types have been purified by dual beam sorting with efficiencies up to 97 [21] Instead of sorting chromosomes isolated from human cells it is also possible to sort human chromosomes isolated from Chinese hamster x human hybrid cells which contain a reduced number of human chromosome types [122223] In a number of cases this allows an increased purification

Improvement of Flow Cytometric Resolution of Specific Chromosome Types by Chemical Modification of DNA

Under favourable circumstances it is nowadays possible to detect quantitatively changes in DNA content as smaIl as 10- 15 g [4]

However there are problems which may requiie addishytional discrimination For example it may be interesting to analyze the chromosomes of other mammals or to improve the flow cytometric analysis of chromosomes isolated from tumor cells which often show a variety of complex rearshyrangements

A) Incorporation of Base Analogues

One possibility to obtain an additional discrimination of specific chromosome types [24 25J is to use the fact that the chromosomal DNA may be replicated at different times of the cell cycle

The Chinese hamster Y chromosome for example is the last to be replicated If the base analogue 5-bromo-2-deoxyshyuridine (BrdUrd) is incorporated instead of thymidine durshying the last hours prior to mitosis then the AT -specific flushyorescence of the dye Hoechst 33258 is quenched I

Fig 6 gives an example [25] Here bivariate contour plots are shown as indicated in Fig5 The only difference is that in this case chromosomes of a Chinese hamster cellline (M3-

11) [10 26] were used instead of human chromosomes Fig 6a presents the dual parameter flow karyotype of these IChinese hamster chromosomes after HOCA3 staining but without any other modification Here the peak of the Y I chromosomes is not weH separated from neighbouring chroshymosome peaks A microscopic analysis of particles sorted from this Y position revealed that only 15 of the sorted chromosomes were identified as Y chromosomes [24J Such I low purities are insufficient to establish for example a chroshymosome specific DNA library [12 23 27J or to perform other studies eg gene assignment [21] In the two contour plots shown in Fig 6b c BrdUrd was added to the culture I medium 25 hrs (Fig 6b) and 35 hrs (Fig 6c) prior to chroshy I mosome isolation (90 min Colcemid treatment for collection I

of mitotic cells) In this case the HO fluorescence of the Y I chromosome (Ya) was increasingly quenched while the CA3 fluorescence remained about the same Due to this it became

332 C Cremer et al Laser in t t middot Applicati~)ns in Flow

11 OhnBrdU

~ bull xS)

67

(bI 2S hn BrdU

bull fmiddot MI I

Va

10M27 (c) 3S hrsBrdU

Mt ~bullbullxS) 5

9shy

4S8 nm eaCltdtlon (CAJ lIuorlllcnce)

Fig6 Effect of a terminal label with BrdUrd on the bivariate flow karshyyotype of HO - CA3 stained chromosomes M 3-1 cells were incushybated for 0 (a) 25 hrs (b) and 35 hrs (c) with BrdUrd before isolation Chromosome assignment was done according to [26J except for the Y which was identified by dual parameter sorting on microscope slides followed by Q-banding analysis Experimental settings were chosen to highlight the chromosome regions of intershyest (Y 10 11 M2) This resulted in chromosomes 1 and 2 being eliminated from the display Experimental conditions were held

constant for all measurements (from [25J)

possible to c1early separate the position of the Ya peakfrom all other chromosomes In case of the condition shown in

fig 6c~the Y chromosome was sorted with relatively high 1gtf~i~ J7 ) ~~lt ~ ~ ~) ~ ~ ~

~~)J~~fY~0bri~etri~ ~etection ~f Chromosomes Following situ Hy degdizatioDin Suspeosion

- ~lC~~ Ii~ gt - gt lt

dB~Urdapproachdiscussed above is one possibility ~pccmc physico-chemica1 DNA modification may be

Iff4A uOcd to improve the middotszligow cytometric discrimination of

certain chromosome types In practice the BrdUrd method appears to be useful for chromosome types with grossly different replication kinetics [25]

In this section a new approach will be described which in principle may be applied to any chromosome type This new technique is based on the modification of the DNA of specific chromosomes by in situ hybridization with specific DNA probes These probes have been chemically modified eg biotinylated by nick translation then they are hybridshyized to the chromosomes This technique requires denaturshyation (usually done by thermal treatment) of the chromoshysomal DNA prior to hybridization (denaturation temperashyture is typically about 73degC) ie the two native DNA strands are physically separated from each other After renaturation a new double stranded DNA is formed with the complementary sequence of the DNA probe

The biotinylated DNA strands hybridized to the chroshymosomal DNA sequences can be detected eg by a specific fluorcscing antibody detection system (fluorescence hybridshyization chromosome painting) or an enzyme detection reaction

While this technique has already being sucessfully develshyoped for the analysis of chromosomes fixed on slides [28 - 35] its application to flow cytometry of chromosomes is still in the very beginning

Since in this case a suspension of isolated chromosomes is required thc entire fluorescence hybridization protocol has to be performed in suspension maintaining the morshyphological stability of the chromosomes A critical step of this method is the denaturation and renaturation of the chromosomes in suspension This process may be monitored by flow cytometry In the results presented in Fig 7 flow karyotypes of the chromosomes of a Chinese hamster cell line are shown shortly before thermal denaturation (Fig 7a) and at different times afterwards (Figs 7b c d) Immediately following denaturation the temperature was lowered(room temperature) so that reannealing of the two strands was possible For flow cytometry the chromosomes were stained with DAPI a fluorescent dye preferably binding to double stranded (Le renatured) DNA

All flow cytometric measurements prescnted in Fig 7 were performed under the identical instrumental set up (fluoresshycence gain etc) aliquots from the same DAPI stained chroshymosome suspension were used Thc chromosomes of the flow karyotypes in Fig 7b c d were additionally stabilized with propidium iodide (PI) This led to q uenching of the DAPI fluorescence (quantum yield about 50) Thus the flow distributions obtained may be compared with each other

Two minutes after denaturation (Fig 7b) the flow distrishybution was very poorly resolved The DAPI fluorescence of aB chromosomes was found to be significantly reduced This suggests that under the conditions used a considerable amount of denaturation took place With increasing time however the original flow karyotype (Fig 7a) gradually reappeared (Figs 7c d)

lt is interesting to note that in the course of the reanshynealing process the fluorescence intensities (the effect of PI taken into consideration) of the individual chromosome peaks regain values alm ost identical to those prior to deshy

-

330 C Cremer et al Laser in utnrnptlv ApplIcatHms in Flow

interval of abscissa values (2k + 1) seemed to fulfil this transshyformation quite wen for flow karyotypes Here it is assumed that the measured curve cohsists of two parts a true smooth curve and the noise values [18] Within the chosen interval the latter are supposed nearly to sum up to zero Then each ordinate value of the new transformed smooth flow histogram is computed by a weightened sum of the k nearest neighbour ordinate values of the original curve The used weighting factors have the form of binominal factors

aj = (2k + l)jj(k + 1- j) (withj = -k +k)

Under the conditions chosen the area of the total flow karshyyotype changes only less than 01 compared to the original measured histogram

In Fig 3b the result of such a smoothing protocol is shown (applied to the data of Fig 3a) Reducing the apparshyent noise a number of true chromosome peaks is highshylighted

In addition the background was quantified and subshytracted In this case the assumption of a ChP-distribution function resulted in a better background subtraction than the exponential assumption often used (J Doumllle et aL manshyuscript in preparation)

Flow histograms thus obtained may be analyzed further According to theory an ideal linear flow karyotype (without noise and background) is composed of Gaussian distribushytions one for the fluorescence values of each different chroshymosome type [4] The areas of these Gaussians correspond to the relative frequencies of the respective chromosome types in the suspension Peak means and coefficients of varshyiation (CV = standard deviation of the peak divided by peak mean) give precise quantitative information on the composition of the flow karyotype To obtain these data ie composition of the single peaks relative and absolute peak area CV etc flow karyotypes may be approximated as a sum of Gaussian distributions

Considering the rise of the logarithmic amplification used in most flow cytometers it is according to our experience also possible to use the same fit procedure for flow karyoshytypes with a logarithmic distribution along the fluorescence axis In this case the area of a peak typical for a chromoshysome in a logarithmic flow karyotype usually differs less than 1 from the area of the theoretical peak resulting from a Gaussian fit Fig 4 gives an example for this kind of evalshyuation In Fig 4a a flow karyotype of the Chinese hamster cellline CHL is presented after application of the smoothing protocol and background subtraction as described above In Fig 4b a theoretical flow karyotype is shown composed of 10 single Gaussian curves by an interactive PC program Peaks consisting of several chromosome types are separated into single overlapping peaks The resulting curve summing up those peaks fits the measured flow karyotype quite weIl (difIerence in peak area less than 2) Data evaluation procedures such as those described

above may contribute considerably to improve the analysis gt~~~()f~xperimental flow karyotypes However the decompositshyllon into Gaussian distributions requires low background

middotijtJ~ve1s and sufiiciently separated peaks to be reliable fiCli7

Fig4 Matching of Gaussians to a flow karyotype

ordinate Number of partic1es abscissa Channel number (correshysponding to the logarithm of DAPI fluorescence intensity) a) flow karyotype (Chinese hamster cell line CHL) after smoothing and background subtraction (see Fig 3) of the experimental data b) decomposition of the flow karyotype (Fig 4a) into 10 Gaussian fit

curves

Since noise and background levels critically depend on the experimental conditions (eg chromosome isolation proshycedures number of particles measured laser light intensity and stability stability of the fluid flow system electronical set up) a careful adjustment of these parameters may be critical for sufficiently weIl resolved flow karyotypes

Under the conditions described so far the peak separashytion however may not be improved beyond a certain limit ie if two chromosome types have the same total DNA content they cannot be separated by univariate flow cytoshymetry alone using DNA stains which do not prefer either AT or GC base pairs (for instance intercalators like Proshypidium Iodide or Ethidium Bromide) Therefore a signifishycant progress in this context was achieved by using AT GC specific DNA stains (eg Hoechst 33258 Hoechst 33342 Chromomycin A3 Mithramycin etc) together with dual lashyser flow cytometry

Dual Laser Flow Cytometry

To perform dual beam (laser) flow cytometry [19] the isolated chromosomes are stained with two fluorescent dyes instead of one only These two dyes have different base speshy

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 331

cifity and different excitation and emission wavelengths An intensively used dye combination is Hoechst 33258 (HO) wruch is AT specific and Chromomycin A3 (CA3) which is GC specific

In a dual laser flow cytometer the stained chromosomes pass one after another two slightly elliptically focused CW laser beams of different wavelengths In case of the fluoroshychromes named above UV multiline excitation is used for HO blue laser light (mainly 458 nm) is used to excite CA3 fluorescence The fluorescence emissions are collected inshydependently from each other This is possible due to the spatial distance of the two laser intersection points Thus information may be obtained not only on the relative DNA content (06 x HO + 04 x CA3 fluorescence intensity) [26J but on the AT (HO fluorescence)GC (CA3 fluoresshycence) ratio as weIl

4)

Fig5 Bivariate distribution (dual laser flow cytometry) of chromosomes isolated from peripheral human lymphocytes After removal of the blood the chromosomes were isolated following 90 min treatment with demecolcine (Colcemid) and then measured with the Livershymore dual laser flow cytometer [19] No noise level or background elimination procedure was applied ordinate HO fluorescence inshytensity (64 channels UV multiline excitation) abscissa CA3 fluoshyrescence intensity (64 channels 458 nm excitation) The fluorescence amplification was linear for both parameters The contour lines of equal particle frequencies are shown (relative levels 1006228 13 6 3 1 05) The peak identification was made according to [20] Due to the fluorescence gain used the large chromosomes No 1 23 were excluded from the display This allowed a better discrimshy

ination of the smaller chromosomes

Fig5 shows an application of dual laser flow cytometry in the quantitative analysis of human chromosomes Here chromosomes were isolated [20J from a short term culture (72 hours) of female peripheral blood (normal donor) The ordinate of this bivariate plot shows the fluorescence of the AT specific dye HO the abscissa presents the fluoresshycence of the GC specific dye CA3 The contour curves inshydicate the lines of equal partic1e frequency Therefore c10sed contour lines represent a weIl separated chromosome peak All of the 20 different chromosome types measured he re (No 4-22 X) may be assigned [20] In this example eleven of

them (chromosome types No 4 5 6 13 16 17 18 19 20- 21 22) are represented by individual chromosome pe~ ~~ Using dual laser flow cytometrymiddot togetherwith the fluor~i~ chromes HO and CA3 it is possible to discriminate all hu-i~ man chromosomes [20] except chromosomes No 9-12 which are too similar both in total DNA content and in ATGC ratio Some further progress in the resolution of the 9 -12 peak is possible using the fluorescence dye DIPI (preshyfering AT base pairs [21]) instead of HO

Using the sorting procedure described above individual human chromosome types have been purified by dual beam sorting with efficiencies up to 97 [21] Instead of sorting chromosomes isolated from human cells it is also possible to sort human chromosomes isolated from Chinese hamster x human hybrid cells which contain a reduced number of human chromosome types [122223] In a number of cases this allows an increased purification

Improvement of Flow Cytometric Resolution of Specific Chromosome Types by Chemical Modification of DNA

Under favourable circumstances it is nowadays possible to detect quantitatively changes in DNA content as smaIl as 10- 15 g [4]

However there are problems which may requiie addishytional discrimination For example it may be interesting to analyze the chromosomes of other mammals or to improve the flow cytometric analysis of chromosomes isolated from tumor cells which often show a variety of complex rearshyrangements

A) Incorporation of Base Analogues

One possibility to obtain an additional discrimination of specific chromosome types [24 25J is to use the fact that the chromosomal DNA may be replicated at different times of the cell cycle

The Chinese hamster Y chromosome for example is the last to be replicated If the base analogue 5-bromo-2-deoxyshyuridine (BrdUrd) is incorporated instead of thymidine durshying the last hours prior to mitosis then the AT -specific flushyorescence of the dye Hoechst 33258 is quenched I

Fig 6 gives an example [25] Here bivariate contour plots are shown as indicated in Fig5 The only difference is that in this case chromosomes of a Chinese hamster cellline (M3-

11) [10 26] were used instead of human chromosomes Fig 6a presents the dual parameter flow karyotype of these IChinese hamster chromosomes after HOCA3 staining but without any other modification Here the peak of the Y I chromosomes is not weH separated from neighbouring chroshymosome peaks A microscopic analysis of particles sorted from this Y position revealed that only 15 of the sorted chromosomes were identified as Y chromosomes [24J Such I low purities are insufficient to establish for example a chroshymosome specific DNA library [12 23 27J or to perform other studies eg gene assignment [21] In the two contour plots shown in Fig 6b c BrdUrd was added to the culture I medium 25 hrs (Fig 6b) and 35 hrs (Fig 6c) prior to chroshy I mosome isolation (90 min Colcemid treatment for collection I

of mitotic cells) In this case the HO fluorescence of the Y I chromosome (Ya) was increasingly quenched while the CA3 fluorescence remained about the same Due to this it became

332 C Cremer et al Laser in t t middot Applicati~)ns in Flow

11 OhnBrdU

~ bull xS)

67

(bI 2S hn BrdU

bull fmiddot MI I

Va

10M27 (c) 3S hrsBrdU

Mt ~bullbullxS) 5

9shy

4S8 nm eaCltdtlon (CAJ lIuorlllcnce)

Fig6 Effect of a terminal label with BrdUrd on the bivariate flow karshyyotype of HO - CA3 stained chromosomes M 3-1 cells were incushybated for 0 (a) 25 hrs (b) and 35 hrs (c) with BrdUrd before isolation Chromosome assignment was done according to [26J except for the Y which was identified by dual parameter sorting on microscope slides followed by Q-banding analysis Experimental settings were chosen to highlight the chromosome regions of intershyest (Y 10 11 M2) This resulted in chromosomes 1 and 2 being eliminated from the display Experimental conditions were held

constant for all measurements (from [25J)

possible to c1early separate the position of the Ya peakfrom all other chromosomes In case of the condition shown in

fig 6c~the Y chromosome was sorted with relatively high 1gtf~i~ J7 ) ~~lt ~ ~ ~) ~ ~ ~

~~)J~~fY~0bri~etri~ ~etection ~f Chromosomes Following situ Hy degdizatioDin Suspeosion

- ~lC~~ Ii~ gt - gt lt

dB~Urdapproachdiscussed above is one possibility ~pccmc physico-chemica1 DNA modification may be

Iff4A uOcd to improve the middotszligow cytometric discrimination of

certain chromosome types In practice the BrdUrd method appears to be useful for chromosome types with grossly different replication kinetics [25]

In this section a new approach will be described which in principle may be applied to any chromosome type This new technique is based on the modification of the DNA of specific chromosomes by in situ hybridization with specific DNA probes These probes have been chemically modified eg biotinylated by nick translation then they are hybridshyized to the chromosomes This technique requires denaturshyation (usually done by thermal treatment) of the chromoshysomal DNA prior to hybridization (denaturation temperashyture is typically about 73degC) ie the two native DNA strands are physically separated from each other After renaturation a new double stranded DNA is formed with the complementary sequence of the DNA probe

The biotinylated DNA strands hybridized to the chroshymosomal DNA sequences can be detected eg by a specific fluorcscing antibody detection system (fluorescence hybridshyization chromosome painting) or an enzyme detection reaction

While this technique has already being sucessfully develshyoped for the analysis of chromosomes fixed on slides [28 - 35] its application to flow cytometry of chromosomes is still in the very beginning

Since in this case a suspension of isolated chromosomes is required thc entire fluorescence hybridization protocol has to be performed in suspension maintaining the morshyphological stability of the chromosomes A critical step of this method is the denaturation and renaturation of the chromosomes in suspension This process may be monitored by flow cytometry In the results presented in Fig 7 flow karyotypes of the chromosomes of a Chinese hamster cell line are shown shortly before thermal denaturation (Fig 7a) and at different times afterwards (Figs 7b c d) Immediately following denaturation the temperature was lowered(room temperature) so that reannealing of the two strands was possible For flow cytometry the chromosomes were stained with DAPI a fluorescent dye preferably binding to double stranded (Le renatured) DNA

All flow cytometric measurements prescnted in Fig 7 were performed under the identical instrumental set up (fluoresshycence gain etc) aliquots from the same DAPI stained chroshymosome suspension were used Thc chromosomes of the flow karyotypes in Fig 7b c d were additionally stabilized with propidium iodide (PI) This led to q uenching of the DAPI fluorescence (quantum yield about 50) Thus the flow distributions obtained may be compared with each other

Two minutes after denaturation (Fig 7b) the flow distrishybution was very poorly resolved The DAPI fluorescence of aB chromosomes was found to be significantly reduced This suggests that under the conditions used a considerable amount of denaturation took place With increasing time however the original flow karyotype (Fig 7a) gradually reappeared (Figs 7c d)

lt is interesting to note that in the course of the reanshynealing process the fluorescence intensities (the effect of PI taken into consideration) of the individual chromosome peaks regain values alm ost identical to those prior to deshy

-

C Cremer et al Laser in Cytometry Applications in Flow Cytogenetics 331

cifity and different excitation and emission wavelengths An intensively used dye combination is Hoechst 33258 (HO) wruch is AT specific and Chromomycin A3 (CA3) which is GC specific

In a dual laser flow cytometer the stained chromosomes pass one after another two slightly elliptically focused CW laser beams of different wavelengths In case of the fluoroshychromes named above UV multiline excitation is used for HO blue laser light (mainly 458 nm) is used to excite CA3 fluorescence The fluorescence emissions are collected inshydependently from each other This is possible due to the spatial distance of the two laser intersection points Thus information may be obtained not only on the relative DNA content (06 x HO + 04 x CA3 fluorescence intensity) [26J but on the AT (HO fluorescence)GC (CA3 fluoresshycence) ratio as weIl

4)

Fig5 Bivariate distribution (dual laser flow cytometry) of chromosomes isolated from peripheral human lymphocytes After removal of the blood the chromosomes were isolated following 90 min treatment with demecolcine (Colcemid) and then measured with the Livershymore dual laser flow cytometer [19] No noise level or background elimination procedure was applied ordinate HO fluorescence inshytensity (64 channels UV multiline excitation) abscissa CA3 fluoshyrescence intensity (64 channels 458 nm excitation) The fluorescence amplification was linear for both parameters The contour lines of equal particle frequencies are shown (relative levels 1006228 13 6 3 1 05) The peak identification was made according to [20] Due to the fluorescence gain used the large chromosomes No 1 23 were excluded from the display This allowed a better discrimshy

ination of the smaller chromosomes

Fig5 shows an application of dual laser flow cytometry in the quantitative analysis of human chromosomes Here chromosomes were isolated [20J from a short term culture (72 hours) of female peripheral blood (normal donor) The ordinate of this bivariate plot shows the fluorescence of the AT specific dye HO the abscissa presents the fluoresshycence of the GC specific dye CA3 The contour curves inshydicate the lines of equal partic1e frequency Therefore c10sed contour lines represent a weIl separated chromosome peak All of the 20 different chromosome types measured he re (No 4-22 X) may be assigned [20] In this example eleven of

them (chromosome types No 4 5 6 13 16 17 18 19 20- 21 22) are represented by individual chromosome pe~ ~~ Using dual laser flow cytometrymiddot togetherwith the fluor~i~ chromes HO and CA3 it is possible to discriminate all hu-i~ man chromosomes [20] except chromosomes No 9-12 which are too similar both in total DNA content and in ATGC ratio Some further progress in the resolution of the 9 -12 peak is possible using the fluorescence dye DIPI (preshyfering AT base pairs [21]) instead of HO

Using the sorting procedure described above individual human chromosome types have been purified by dual beam sorting with efficiencies up to 97 [21] Instead of sorting chromosomes isolated from human cells it is also possible to sort human chromosomes isolated from Chinese hamster x human hybrid cells which contain a reduced number of human chromosome types [122223] In a number of cases this allows an increased purification

Improvement of Flow Cytometric Resolution of Specific Chromosome Types by Chemical Modification of DNA

Under favourable circumstances it is nowadays possible to detect quantitatively changes in DNA content as smaIl as 10- 15 g [4]

However there are problems which may requiie addishytional discrimination For example it may be interesting to analyze the chromosomes of other mammals or to improve the flow cytometric analysis of chromosomes isolated from tumor cells which often show a variety of complex rearshyrangements

A) Incorporation of Base Analogues

One possibility to obtain an additional discrimination of specific chromosome types [24 25J is to use the fact that the chromosomal DNA may be replicated at different times of the cell cycle

The Chinese hamster Y chromosome for example is the last to be replicated If the base analogue 5-bromo-2-deoxyshyuridine (BrdUrd) is incorporated instead of thymidine durshying the last hours prior to mitosis then the AT -specific flushyorescence of the dye Hoechst 33258 is quenched I

Fig 6 gives an example [25] Here bivariate contour plots are shown as indicated in Fig5 The only difference is that in this case chromosomes of a Chinese hamster cellline (M3-

11) [10 26] were used instead of human chromosomes Fig 6a presents the dual parameter flow karyotype of these IChinese hamster chromosomes after HOCA3 staining but without any other modification Here the peak of the Y I chromosomes is not weH separated from neighbouring chroshymosome peaks A microscopic analysis of particles sorted from this Y position revealed that only 15 of the sorted chromosomes were identified as Y chromosomes [24J Such I low purities are insufficient to establish for example a chroshymosome specific DNA library [12 23 27J or to perform other studies eg gene assignment [21] In the two contour plots shown in Fig 6b c BrdUrd was added to the culture I medium 25 hrs (Fig 6b) and 35 hrs (Fig 6c) prior to chroshy I mosome isolation (90 min Colcemid treatment for collection I

of mitotic cells) In this case the HO fluorescence of the Y I chromosome (Ya) was increasingly quenched while the CA3 fluorescence remained about the same Due to this it became

332 C Cremer et al Laser in t t middot Applicati~)ns in Flow

11 OhnBrdU

~ bull xS)

67

(bI 2S hn BrdU

bull fmiddot MI I

Va

10M27 (c) 3S hrsBrdU

Mt ~bullbullxS) 5

9shy

4S8 nm eaCltdtlon (CAJ lIuorlllcnce)

Fig6 Effect of a terminal label with BrdUrd on the bivariate flow karshyyotype of HO - CA3 stained chromosomes M 3-1 cells were incushybated for 0 (a) 25 hrs (b) and 35 hrs (c) with BrdUrd before isolation Chromosome assignment was done according to [26J except for the Y which was identified by dual parameter sorting on microscope slides followed by Q-banding analysis Experimental settings were chosen to highlight the chromosome regions of intershyest (Y 10 11 M2) This resulted in chromosomes 1 and 2 being eliminated from the display Experimental conditions were held

constant for all measurements (from [25J)

possible to c1early separate the position of the Ya peakfrom all other chromosomes In case of the condition shown in

fig 6c~the Y chromosome was sorted with relatively high 1gtf~i~ J7 ) ~~lt ~ ~ ~) ~ ~ ~

~~)J~~fY~0bri~etri~ ~etection ~f Chromosomes Following situ Hy degdizatioDin Suspeosion

- ~lC~~ Ii~ gt - gt lt

dB~Urdapproachdiscussed above is one possibility ~pccmc physico-chemica1 DNA modification may be

Iff4A uOcd to improve the middotszligow cytometric discrimination of

certain chromosome types In practice the BrdUrd method appears to be useful for chromosome types with grossly different replication kinetics [25]

In this section a new approach will be described which in principle may be applied to any chromosome type This new technique is based on the modification of the DNA of specific chromosomes by in situ hybridization with specific DNA probes These probes have been chemically modified eg biotinylated by nick translation then they are hybridshyized to the chromosomes This technique requires denaturshyation (usually done by thermal treatment) of the chromoshysomal DNA prior to hybridization (denaturation temperashyture is typically about 73degC) ie the two native DNA strands are physically separated from each other After renaturation a new double stranded DNA is formed with the complementary sequence of the DNA probe

The biotinylated DNA strands hybridized to the chroshymosomal DNA sequences can be detected eg by a specific fluorcscing antibody detection system (fluorescence hybridshyization chromosome painting) or an enzyme detection reaction

While this technique has already being sucessfully develshyoped for the analysis of chromosomes fixed on slides [28 - 35] its application to flow cytometry of chromosomes is still in the very beginning

Since in this case a suspension of isolated chromosomes is required thc entire fluorescence hybridization protocol has to be performed in suspension maintaining the morshyphological stability of the chromosomes A critical step of this method is the denaturation and renaturation of the chromosomes in suspension This process may be monitored by flow cytometry In the results presented in Fig 7 flow karyotypes of the chromosomes of a Chinese hamster cell line are shown shortly before thermal denaturation (Fig 7a) and at different times afterwards (Figs 7b c d) Immediately following denaturation the temperature was lowered(room temperature) so that reannealing of the two strands was possible For flow cytometry the chromosomes were stained with DAPI a fluorescent dye preferably binding to double stranded (Le renatured) DNA

All flow cytometric measurements prescnted in Fig 7 were performed under the identical instrumental set up (fluoresshycence gain etc) aliquots from the same DAPI stained chroshymosome suspension were used Thc chromosomes of the flow karyotypes in Fig 7b c d were additionally stabilized with propidium iodide (PI) This led to q uenching of the DAPI fluorescence (quantum yield about 50) Thus the flow distributions obtained may be compared with each other

Two minutes after denaturation (Fig 7b) the flow distrishybution was very poorly resolved The DAPI fluorescence of aB chromosomes was found to be significantly reduced This suggests that under the conditions used a considerable amount of denaturation took place With increasing time however the original flow karyotype (Fig 7a) gradually reappeared (Figs 7c d)

lt is interesting to note that in the course of the reanshynealing process the fluorescence intensities (the effect of PI taken into consideration) of the individual chromosome peaks regain values alm ost identical to those prior to deshy

-

332 C Cremer et al Laser in t t middot Applicati~)ns in Flow

11 OhnBrdU

~ bull xS)

67

(bI 2S hn BrdU

bull fmiddot MI I

Va

10M27 (c) 3S hrsBrdU

Mt ~bullbullxS) 5

9shy

4S8 nm eaCltdtlon (CAJ lIuorlllcnce)

Fig6 Effect of a terminal label with BrdUrd on the bivariate flow karshyyotype of HO - CA3 stained chromosomes M 3-1 cells were incushybated for 0 (a) 25 hrs (b) and 35 hrs (c) with BrdUrd before isolation Chromosome assignment was done according to [26J except for the Y which was identified by dual parameter sorting on microscope slides followed by Q-banding analysis Experimental settings were chosen to highlight the chromosome regions of intershyest (Y 10 11 M2) This resulted in chromosomes 1 and 2 being eliminated from the display Experimental conditions were held

constant for all measurements (from [25J)

possible to c1early separate the position of the Ya peakfrom all other chromosomes In case of the condition shown in

fig 6c~the Y chromosome was sorted with relatively high 1gtf~i~ J7 ) ~~lt ~ ~ ~) ~ ~ ~

~~)J~~fY~0bri~etri~ ~etection ~f Chromosomes Following situ Hy degdizatioDin Suspeosion

- ~lC~~ Ii~ gt - gt lt

dB~Urdapproachdiscussed above is one possibility ~pccmc physico-chemica1 DNA modification may be

Iff4A uOcd to improve the middotszligow cytometric discrimination of

certain chromosome types In practice the BrdUrd method appears to be useful for chromosome types with grossly different replication kinetics [25]

In this section a new approach will be described which in principle may be applied to any chromosome type This new technique is based on the modification of the DNA of specific chromosomes by in situ hybridization with specific DNA probes These probes have been chemically modified eg biotinylated by nick translation then they are hybridshyized to the chromosomes This technique requires denaturshyation (usually done by thermal treatment) of the chromoshysomal DNA prior to hybridization (denaturation temperashyture is typically about 73degC) ie the two native DNA strands are physically separated from each other After renaturation a new double stranded DNA is formed with the complementary sequence of the DNA probe

The biotinylated DNA strands hybridized to the chroshymosomal DNA sequences can be detected eg by a specific fluorcscing antibody detection system (fluorescence hybridshyization chromosome painting) or an enzyme detection reaction

While this technique has already being sucessfully develshyoped for the analysis of chromosomes fixed on slides [28 - 35] its application to flow cytometry of chromosomes is still in the very beginning

Since in this case a suspension of isolated chromosomes is required thc entire fluorescence hybridization protocol has to be performed in suspension maintaining the morshyphological stability of the chromosomes A critical step of this method is the denaturation and renaturation of the chromosomes in suspension This process may be monitored by flow cytometry In the results presented in Fig 7 flow karyotypes of the chromosomes of a Chinese hamster cell line are shown shortly before thermal denaturation (Fig 7a) and at different times afterwards (Figs 7b c d) Immediately following denaturation the temperature was lowered(room temperature) so that reannealing of the two strands was possible For flow cytometry the chromosomes were stained with DAPI a fluorescent dye preferably binding to double stranded (Le renatured) DNA

All flow cytometric measurements prescnted in Fig 7 were performed under the identical instrumental set up (fluoresshycence gain etc) aliquots from the same DAPI stained chroshymosome suspension were used Thc chromosomes of the flow karyotypes in Fig 7b c d were additionally stabilized with propidium iodide (PI) This led to q uenching of the DAPI fluorescence (quantum yield about 50) Thus the flow distributions obtained may be compared with each other

Two minutes after denaturation (Fig 7b) the flow distrishybution was very poorly resolved The DAPI fluorescence of aB chromosomes was found to be significantly reduced This suggests that under the conditions used a considerable amount of denaturation took place With increasing time however the original flow karyotype (Fig 7a) gradually reappeared (Figs 7c d)

lt is interesting to note that in the course of the reanshynealing process the fluorescence intensities (the effect of PI taken into consideration) of the individual chromosome peaks regain values alm ost identical to those prior to deshy

-

C Cremer ct al Laser in Cytometry AppUcations in Flow Cytogcnetics 333

lU

83

1amp1 242

2712B8

Z16

144

Fig7 Flow cytometric monitoringof the denaturationreannealing of chromosomes in suspension (univariate measurements)

a) flow distribution prior to thermal denaturation (DAPI staining) b)-d) flow distributions (DAPI staining) of the same chromosome suspension as shown in a) but additionally stabilized with propidium iodide measured at different reannealing times performed at 2 min (b) 15 min (c) 25 min (d) following denaturation In all cases the

same instrumental set up was used

naturation (data not shown) This suggests that the 10ss of DNA due to denaturation was low under the conditions used

Recently protocols have been described for fluorescence hybridization of chromosomes in suspension which result in a large number of morphologically intact chromosomes as characterized by microscopic analysis [36 - 38] In addition first flow cytometric measurements have been performed [37 39 40]

In Fig 8 flow karyotypes of a Chinese hamster x human hybrid cell line are shown The chromosomes were isolated according to a modified hexandiol method [15J hybridized in suspension with total genomic human DNA and countershytained with DAPL Fig 8a shows the DAPI flow karyotype of this cell line prior to the hybridization procedure while in Fig 8b the karyotype is shown afterwards Compared to Fig 8a the number of events in Fig 8b decreased from sm all to large chromosomes This indicates that the percentage of morphologically intact chromosomes measured by flow cytometry after the hybridization procedure was reduced with increasing chromosomal length~ However the overall structure of the flow karyotype is still recognizable The comparison of the peak position of the different chromoshy

some types indicates that a considerable percentage of the chromosomes has still about the same DAPI fluorescence before and after the hybridization treatment This me ans that renaturation was complete as far as it was measureable by flow cytometry and that the DNA loss was low

From the results presented in Fig 8 the possibility cannot be exc1uded that only the Chinese hamster chromosomes maintllined their morphological stability while the human chromosomes were destabilized Recent experiments [40J however using one and two parameter slit scan flow cytoshymetry (here the laser beam is focused to such an extent that morphological details of the chromosomes become detectshyable [41 42J) suggest that at least a significant part of the human chromosomes remains sufficiently intact to be measshyured by flow cytometry after denaturation hybridization and fluorescence visualization

Future Directions

Slit scan flow cytometry allows to analyze morphologishycally weIl preserved normal and aberrant chromosomes like dicentrics [42J or translocations [3940] Nonetheless an improved application of zero-resolution flow cytometry

334 C Cremer et al Laser in CVtorrletfV Appjicatiolls in Flow

il

Fig8 Flow karyotypes (from [40 47]) of DAPI stained chromosomes of a Chinese hamster x human hybrid cellline A1wbf2 [36-38] before (a) and after (b) fluorescence hybridization and stabilizatfon with ethanol Ordinate Number of partic1es abscissa Channel number (0-64) proportional to the DAPI fluorescence intensity Backshy

ground has been subtraeted by an exponential function

(Fig 1) remains a significant challenge because of its rapid -analysis and sorting capacity Of particular interest in this context is the zero-resolution detection of chromosomes folshylowing fluorescence hybridization in suspension If the techshynical problems ofthis new method (eg hybridization chemshyistry morphological stability of chromosomes) can be solved the rapid characterization (up to several thousand partic1es per second) of the chromosomal constitution of a variety of cell lines might be improved considerably by means of dual parameter zero-resolution flow cytometry It is important to note that the range of potential flow applishycations (both zero-resolution and one dimensional ie slit scan flow cytometry) will critically depend on the DNA probes available for fluorescence hybridization Ideally DNA probes are required to highlight specifically any chroshymosome or chromosome region especially of the human complement To realize this goal recently significant proshygress has been achieved to delineate chromosomes in human rells by fluorescence hybridization with high specificity [32 35 43 44] These methods are mainly based on the use of

chromosome specific DNA libraries established from flow tsorted chromosomes [11 12 23 27] So far it has been shown that these delineation procedures allow a considershy~ble improvement of the cytological analysis of cells inc1udshy

ing tumor cells [3244] With further developments ofmultishyprobe hybridization and detection procedures [4546] these methods might eventually also be applied in flow cytometry to a rapid screening of tumor cell populations F or instance using two colour fluorescence hybridization [31 - 33] it might become possible to realize the following approach

Chromosomes (eg from a tumor cell population) are isoshylated chromosomes of type Aare hybridized and labelled for green fluorescence while chromosomes of type Bare hybridized and labelIed for red fluorescence In a bivariate contourplot (eg abscissa green fluorescence intensity orshydinate red fluorescence intensity) chromosomes of type A should form a peak near the abscissa chromosomes of type B should form a peak near the ordinate while translocation chromosomes (exchange of material between chromosome types A and B) should form peaks in between

As compared to the flow cytometric analysis of chromoshysome translocations with conventional DNA stains [4] such a method should allow to highlight specifically the flow disshytributions of the chromosomes of interest in the particular case (e g in leukemias)

Fig 1 was obtained by courtesy of the Coulter Corporation The experimental flow distributions for Figs 3 4 8 were measured by M H using the flow cytometer at the University of Tuumlbingen (Dr H-J Buumlhring) The data presented in Fig 5 were measured by e e during his stay at the Lawrence Livermore National Laboratory (Dr J W Gray) Furthermore we thank Dr T Cremer (Yale UniversityjHeidelberg University) for the oommunication of unshypublished manuseripts concerning the specific delineation ofhuman chromosomes

Tbe support of the Deutsche Forschungsgemeinschaft is grateshyfully acknowledged

References

[1] D Peters E Branscomb P Dean T Merril D Pinkel M A Van Dilla and J W Gray Cytometry 6 290 (1985)

[2] B Barlogie M N Raber J Schumann T S Johnson B Drewinlo D E Swartzendruber W Goumlhde M Andruff and E Freireieh Cancer Res 43 3982 (1983)

[3] O S Frankfurt H U Slocum Y M Rostim S G Arbruck Z P Pavelic N Petrelli R P Huber E J Pontes and W R Greeo Cytometry 5 71 (1984)

[4] J W Gray and R G Langlois Ann Rev Biophys Biophys Chem 15 195 (1986)

[5] M A Van Dilla T T Trujillo P F Mullaney and J R Coulter Seience 163 1213 (1969)

[6] P K Horan and L L Wheeless Jr Science 198 149 (1977) [7] M Melamed P Mullaney and M L Mendelsohn (Eds)

Flow cytometry and sorting p 716 J Wiley New York 1979 [8] M A Van Dilla P N Dean O Laerum and M Melamed

(Eds) Flow cytometry Instrumentation and data analysis p 288 Academic Press New Y ork 1985

[9] J V Watson Nature 325 741 (1987) [10] J W Gray A V Carrano L L Steinmetz M A Van Dilla

D H Moore H B H Meyall and M L Mendelsohn Proc Nat Acad Sei USA 72 1231 (1975)

[11] J e Fuscoe L M Clark and M A Van Dilla Cytogen Cell Gen 43 79 (1986)

[12] M A Van Dilla et al Bioteehnology 4 537 (1986) [13] L-e Yu J A Aten J W Gray and A V Carrano Nature

293154 (1981) [14] M Stoumlhr K J Hutter M Frank and K Goerttler Histoshy

chemistry 74 57 (1982) [15] J Barths Dissertation Universitaumlt Kaiserslautern (FRG)

1987 [16] e Cremer M Hausmann P Zuse J A Aten J Barths and

H-J Buumlhring Optik (submitted) (1988)

S Andersson-Engels et al Tissue Diagnostics Using Laser-Induced Fluorescence

[17] C Cremer T Cremer and 1 W Gray Cytometry 2 287 (1982)

[18] J Hartung Statistik p200 Oldenbourg Verlag Muumlnchen 1984

[19] P N Dean and D C Pinkel J Histochem 26 622 (1978) [20] R G Langlois L-c Yu 1 W Gray and A V Carrano

Proe Natl Aead Sei USA 79 7876 (1982) [21] R V Lebo F Gorin R J Fletterie F-T Kao M C Cheung

B D Bruce and Y W Kan Science 225 57 ~1984) [22J C Cremer 1 W Gray and H H Ropers Hum Genet 60

262 (1982) [23J C Cremer G RappoId J W Gray C R Muumlller and H H

Ropers Cytometry 5 572 (1984) [24] C Cremer and J W Gray Somat Cell Genet 8 319 (1982) [25J C Cremer and 1 W Gray Cytometry 3 282 (~83) [26J J W Gray R G Langlois A V Carrano K J Burkhartshy

Schultz and M A Van DilJa Chromosoma (Berl) 73 9 (1979)

[27J K E Davies B D Young R G ElIes M E Hill and R Williamson Nature 293 374 (1981)

[28J M Schardin T Cremer H D Hager and M Lang Hum Genet 71281 (1985)

[29] T Cremer 1 Landegent A Bruumlckner H P Scholl M Scharshydin H D Hager P Devilee P Pearson and M van der Ploeg Hum Genet 74 346 (1986)

[30] D Pinkel T Straume and J W Gray Proc Nat Acad Sei (USA) 83 2934 (1986)

[31] T Cremer D Tesin A H N Hopman and L Manuelidis Exp Cell Res 176 199 (1988)

[32J T Cremer P Lichter J Borden D C Ward and L Manshyuelidis Hum Genet 80 235 (1988)

[33] P Emmerich P Loos A Jauch A H N Hopman J Wiegant M Higgins B N White M van der Ploeg C Creshymer and T Cremer Exp Cell Res (in press) (1989)

[34] C Cremer M Hausmann E Diaz J Hetzel J A Aten and T Cremer in Automation of Cytogenetics - Advances in Systems and Techniques eds C Lundsteen and J Piper Springer Heidelberg 1989 (in press)

[35] P Lichter T CremerJBor~t Manuelidis andD~C Ward Hum Genet 80 224 (1988)Jmiddot~ ~J ) At~~Lbull

[36] G Dudin T Cremer MSeh8rdinM Hausmann ~Bi~t and C Cremer Hum Genet 76290 (1987) jJ~

[37] G Dudin M Hausmann W Rens J A Aten and CCremet Ann Univ Sarav Med Suppl 7 81 (1987)

[38] G Dudin E W Steegmayer P Vogt H Schnitzer E Diaz K Howell T Cremer and C Cremer Hum Genet 80 111 (1988)

[39] M Hausmann G Dudin J A Aten F Bier and C Cremer Eur J Cell Biol (Suppl 22) 46 26 (1988)

[40J M Hausmann G Dudin 1 A Aten H-1 Buumlhring E Diaz J Doumllle F Bier and C Cremer Biomed Optics (in press) (1989)

[41] 1 W Gray D Peters 1 T Merrill R Martin and M A van Dilla J Histochem Cytochem 27 441 (1979)

[42J 1 N Lucas and 1 W Gray Cytometry 8 273 (1987) [43J D Pinkel J Lundegent C CoHins R Segraves1 Lucas and

1 Gray Proc NatL Acad Sei (USA) 85 9138 (1988) [44J P Lichter T Cremer C Chang Tang P C Watkins L

Manuelidis and D C Ward Proc Natl Acad Sei (USA) 85 9664 (1988)

[45J P M Nederlof D Robinson 1 Wiegant A H N Hopman H 1 Tanke and A K Raap Cytometry Supp 2 1988 80 (1988)

[46] F M Waldman D Pinkel J Mullikin and J W Gray Cyshytometry Supp 2 1988 81 (1988)

[47] M Hausmann Dissertation Fakultaumlt fuumlr Physik und Astroshynomie Universitaumlt Heidelberg 1988

[48) F F Bier U Bettag T Rheingans Hmiddot Adrian 1 Barths M Hausmann G Dudin H-J Buumlhring P Rohwer J Doumllle and C Cremer Electrophoresis (submitted) (1988)

Presented at the Discussion Meeting ofthe E 6933 Deutsche Bunsen-Gesellschaft fuumlr Physishykalische Chemie Laser in Life Seiences University of Heidelberg August 29th to September 1st 1988

S Andersson-Engels et al Tissue Diagnostics Using Laser-Induced Fluorescence

[17] C Cremer T Cremer and 1 W Gray Cytometry 2 287 (1982)

[18] J Hartung Statistik p200 Oldenbourg Verlag Muumlnchen 1984

[19] P N Dean and D C Pinkel J Histochem 26 622 (1978) [20] R G Langlois L-c Yu 1 W Gray and A V Carrano

Proe Natl Aead Sei USA 79 7876 (1982) [21] R V Lebo F Gorin R J Fletterie F-T Kao M C Cheung

B D Bruce and Y W Kan Science 225 57 ~1984) [22J C Cremer 1 W Gray and H H Ropers Hum Genet 60

262 (1982) [23J C Cremer G RappoId J W Gray C R Muumlller and H H

Ropers Cytometry 5 572 (1984) [24] C Cremer and J W Gray Somat Cell Genet 8 319 (1982) [25J C Cremer and 1 W Gray Cytometry 3 282 (~83) [26J J W Gray R G Langlois A V Carrano K J Burkhartshy

Schultz and M A Van DilJa Chromosoma (Berl) 73 9 (1979)

[27J K E Davies B D Young R G ElIes M E Hill and R Williamson Nature 293 374 (1981)

[28J M Schardin T Cremer H D Hager and M Lang Hum Genet 71281 (1985)

[29] T Cremer 1 Landegent A Bruumlckner H P Scholl M Scharshydin H D Hager P Devilee P Pearson and M van der Ploeg Hum Genet 74 346 (1986)

[30] D Pinkel T Straume and J W Gray Proc Nat Acad Sei (USA) 83 2934 (1986)

[31] T Cremer D Tesin A H N Hopman and L Manuelidis Exp Cell Res 176 199 (1988)

[32J T Cremer P Lichter J Borden D C Ward and L Manshyuelidis Hum Genet 80 235 (1988)

[33] P Emmerich P Loos A Jauch A H N Hopman J Wiegant M Higgins B N White M van der Ploeg C Creshymer and T Cremer Exp Cell Res (in press) (1989)

[34] C Cremer M Hausmann E Diaz J Hetzel J A Aten and T Cremer in Automation of Cytogenetics - Advances in Systems and Techniques eds C Lundsteen and J Piper Springer Heidelberg 1989 (in press)

[35] P Lichter T CremerJBor~t Manuelidis andD~C Ward Hum Genet 80 224 (1988)Jmiddot~ ~J ) At~~Lbull

[36] G Dudin T Cremer MSeh8rdinM Hausmann ~Bi~t and C Cremer Hum Genet 76290 (1987) jJ~

[37] G Dudin M Hausmann W Rens J A Aten and CCremet Ann Univ Sarav Med Suppl 7 81 (1987)

[38] G Dudin E W Steegmayer P Vogt H Schnitzer E Diaz K Howell T Cremer and C Cremer Hum Genet 80 111 (1988)

[39] M Hausmann G Dudin J A Aten F Bier and C Cremer Eur J Cell Biol (Suppl 22) 46 26 (1988)

[40J M Hausmann G Dudin 1 A Aten H-1 Buumlhring E Diaz J Doumllle F Bier and C Cremer Biomed Optics (in press) (1989)

[41] 1 W Gray D Peters 1 T Merrill R Martin and M A van Dilla J Histochem Cytochem 27 441 (1979)

[42J 1 N Lucas and 1 W Gray Cytometry 8 273 (1987) [43J D Pinkel J Lundegent C CoHins R Segraves1 Lucas and

1 Gray Proc NatL Acad Sei (USA) 85 9138 (1988) [44J P Lichter T Cremer C Chang Tang P C Watkins L

Manuelidis and D C Ward Proc Natl Acad Sei (USA) 85 9664 (1988)

[45J P M Nederlof D Robinson 1 Wiegant A H N Hopman H 1 Tanke and A K Raap Cytometry Supp 2 1988 80 (1988)

[46] F M Waldman D Pinkel J Mullikin and J W Gray Cyshytometry Supp 2 1988 81 (1988)

[47] M Hausmann Dissertation Fakultaumlt fuumlr Physik und Astroshynomie Universitaumlt Heidelberg 1988

[48) F F Bier U Bettag T Rheingans Hmiddot Adrian 1 Barths M Hausmann G Dudin H-J Buumlhring P Rohwer J Doumllle and C Cremer Electrophoresis (submitted) (1988)

Presented at the Discussion Meeting ofthe E 6933 Deutsche Bunsen-Gesellschaft fuumlr Physishykalische Chemie Laser in Life Seiences University of Heidelberg August 29th to September 1st 1988