- Email: [email protected]
The proximity of DNA sequences in interphase cell nuclei is correlated to genomic distance and permits ordering of cosmids spanning 250 kilobase pairs
GENOMICS
5, 710-717
(1989)
The Proximity of DNA Sequences in Interphase Cell Nuclei Is Correlated to Genomic Distance and Permits Ordering of Cosmids Spanning 250 Kilobase Pairs’ BARBARA TRASK,’ Biomedical
Sciences Divkon,
Lawrence
DAN PINKEL, AND GER VAN DEN ENGH
LIvermore National Laboratory, Received
April
5, 1989,
revised
Inc.
INTRODUCTION The concept of interphase chromatin mapping to determine the relative distance between two sequences in the genome was recently introduced by Lawrence et al. (1988). The technique relies on specific labeling of single copy target sequences in interphase nuclei with ’ The U.S. Government’s right to retain a nonexclusive free license in and to the copyright covering this paper, mental purposes, is acknowledged. * To whom correspondence should he addressed.
0888.7543/89
Copyright All rights
$3.00
8 1989 by Academic Press. Inc. of reproduction in any form reserved.
15.
California
94550
1989
fluorescence in situ hybridization (FISH). With FISH, complementary DNA sequence probes are hybridized to nuclei and are subsequently detected with a fluorescent label, resulting in a sharply defined fluorescent spot at the site of hybridization. Lawrence et al. (1988) determined the orientation of two integrated viral genomes from the intensity and spacing of the fluorescent spots obtained by hybridizing DNA sequence probes of different portions of the EpsteinBarr virus t.o nuclei from infected cells. They assumed the relationship bet.ween genomic distance and interphase distance to be linear with zero-intercept to conclude that the viral sequences were separated by 340 kb and to estimate a packing order of 100 for interphase chromatin. This work led to the hope that. analysis of probe separation in interphase could be used to map more closely spaced sequences and could be developed into a tool for genome ordering studies. For this application, the relationship between interphase distance (in pm) and genomic distance (in kb) over a range of genomic distances must be determined. A set of cosmids containing fragments of the Chinese hamst,er (CHO) genome including and surrounding the dihydrofolate reductase (DHFR) gene has been isolat,ed and characterized by Hamlin and co-workers (Looney and Hamlin, 1987; Ma et al., 1988).. This set of cosmids provides an opportunit,y to study the relationship of physical distance between hybridization sites in interphase nuclei to known genomic distance between fragments over a range of 25 to 250 kb. Cells carrying various chromosome inversions that disrupt the DHFR gene were used to further analyze interphase distance between sequences separated by 50-90 Mb.
The physical distance between DNA sequences in interphase nuclei was determined using eight cosmids containing fragments of the Chinese hamster genome that span 273 kb surrounding the dihydrofolate reductase (DHFR) gene. The distance between these sequences at the molecular level has been determined previously by restriction enzyme mapping (J. E. Looney and J. L. Hamlin, 1987, Mol. Cell Biol. 7: 569577; C. Ma et al., 1988, Mol. Cell Biol. 8: 2316-2327). Fluorescence in situ hybridization was used to localize the DNA sequences in interphase nuclei of cells bearing only one copy of this genomic region. The distance between DNA sequences in interphase nuclei was correlated to molecular distance over a range of 25 to at least 250 kb. The observed relationship was such that genomic distance could be predicted to within 40 kb from interphase distance. The correct order of seven probes was derived from interphase distances measured for 19 pair-wise combinations of the probes. Measured distances between sequences = 200 kb apart indicate that the DNA is condensed 70- to loo-fold in hybridized nuclei relative to a linear DNA helix molecule. Cell lines with chromosome inversions were used to show that interphase distance increases with genomic distance in the 50-90 Mb range, but less 8~ 1989 Academic steeply than in the 25-250 kb range. Press.
June
P.O. Box 5507 L-452, Livermore,
EXPERIMENTAL
METHODS
Cell Lines
royaltyfor govern-
1JA41 is a y-ray mutant in which the DHFR gene on chromosome 2 has been deleted leaving only one copy on chromosome 22 (Urlaub et al., 1983). DG46, 710
SEQUENCE
PROXIMITY
IN
DG24, and DG23 are DHFR-deficient mutants, obtained after y-ray treatment of UA41, in which the DHFR locus is disrupted by inversion in metaphase chromosomes. The inversions in DG46 (inv(Z2) dup(Z2) (pter -+ ~23: :p14 -W p23::p14 -+ q37: :q24 q37)) and DG24 (inv(Z2)(p23p31)) were detected by banding (Urlaub et al., 1986) and by FISH to metaphase chromosomes (B. J. Trask, unpublished observations); the inversion in DG23 (inv(Z2) (~23~12)) was detected by FISH to metaphase chromosomes (ibid.). The distance in metaphase between breakpoints in the inversion mutants was measured as a fraction of the length of the 22 chromosome. The DNA content of the inverted segment then was calculated using an estimate oft he DNA content of 22 (390 Mb), based on the DNA content of CHO chromosome 2 (Funanage and Myoda, 1986) and the relative idiogram lengths of chromosomes 22 and 2 (Deaven and Peterson, 1973). Slide Preparation Interphase cells were harvested by trypsinization after 2-4 days growth at complete confluency, incubated 15 min in 75 mMKC1 at 37”C, fixed in methanol:acetic acid (3:1), and dropped on slides. Several observations led us to consider these preparations to contain a relatively pure population of cells in G, : (i) the size of all nuclei was approximately the same, suggesting a low frequency of cells in G,; (ii) no mitotic cells were observed on these slides; and (iii) few cells were observed to be labeled with two fluorescent spots when single probes were hybridized (see Results). Metaphase spreads were prepared from cells collected by mitotic shake-off from nonconfluent cultures incubated 1.5 h in medium containing 0.1 pg/ml colcemid. Slides were baked 4 h at 65°C to artificially age them and stored at -20°C in N:!-enriched atmosphere until use. Probes and Hybridization Purified cosmid DNA was biotinylated by nick translation with biotin-ll-dUTP (Bethesda Research Laboratories). Probe modification averaged 35%. Probes were hybridized to cells using a modification of a published procedure (Pinkel et al.. 1986). Briefly, slides were treated with RNase (100 pg/ml in 2X SSC) prior to being denatured for 2 min at 70°C in 70% formamide/2x SSC and then quenched in cold 70% ethanol. After dehydration, 10 ~1 of a hybridization mixture containing 1-2 ng/pl probe or pooled probes, 50% formamide, 2X SSC, 10% dextran sulfate, and 1 mg/ml Chinese hamster genomic DNA (purified (Maniatis et al., 1982) from AA8 cells, sonicated to 200-600 bp, and resuspended after ethanol precipitation at 10 mg/ml in 1X TE) was applied to slides. Slides were incubated overnight and then washed in 50% formamide/2X SSC and in 2~ SSC. Avidin conjugated
INTERPHASE
CELL
711
NIICLEI
to fluorescein (DCS, Vector Laboratories) was applied to the slides for 30 min at 37°C in PNM buffer (0.1 M NaH2P04, 0.1 M Na2HP04, 0.1% NP-40, 5% nonfat dry milk, pH 8) or in 4X SSC containing 1% BSA (Lawrence et al., 1988). Free avidin was removed by washing slides in several changes of PNM buffer without milk or in 4~ SSC/l% Triton X-100, depending on the buffer used for avidin incubation. Distance
Analysis
Cells were counterstained and viewed in DAPI present at 1 pg/ml in a fluorescence antifade solution (Johnson and Nogueria, 1981). Color photographs (Kodak Ektachrome 400 t,ransparency film, 90 s exposure) were made of a random selection of at least 20 hybridized interphase nuclei for each cosmid pair using a Zeiss Axiophot microscope equipped with a Zeiss Plan Neofluor oil immersion objective (100X, 1.3 n.a.) and 1.25X optivar (epifluorescence filters for fluorescein: excitation BP485 nm, reflector 510 nm, emission LP 520 nm). Photographic images were projected on a wall, and the distance between the centers of fluorescent spots was determined as described in text. Figures are black and white renditions of color images. RESULTS
The relative position of the eight genomic fragments used in this study has been determined by restriction enzyme mapping (Looney and Hamlin, 1987; Ma et al., 1988) and is shown in Fig. 1. The fragments were cloned from a methotrexate-resistant CHO line with multiple copies of the DHFR gene and flanking sequences. The 273 kb shown in Fig. 1 represent the repeated unit, or amplicon. One cosmid, pA36, contains the junction between amplicons and, therefore, sequences separated by 250 kb in the parental genome (Ma et al., 1988). Cosmids were biotinylated by nick translation and hybridized singly or in pairs to metaphase and interphase cells fixed on slides. Hybridization of repetitive sequences in cosmid probes to target cells was blocked by the addition of sonicated CHO genomic DNA to the hybridization solution (Landegent et al., 1987). The hybridization sites were visualized by a single incubation of slides in fluorescein-conjugated avidin. UA41, a CHO cell line with only one copy of the DHFR region sequences (Urlaub et al., 1983), was chosen as hybridization target to simplify analysis of hybridization sites. Hybridization of two probes, 32Al and SE24, to the metaphase cell in Fig. 2A demonstrates the single chromosomal site for these sequences in UA41. Hybridization of a single cosmid to UA41 interphase nuclei (Fig. 2s) illustrates the small (0.30.6 pm diameter) fluorescent hybridization site obtained with FISH. A low amount of hybridization of repetitive elements in the probe is useful because nu-
712
TRASK, 0
50
N27
PINKEL,
VAN
DEN
150
DHFR+
100
32Al
I
‘26A31 t
FIG. 1. Relative spacing gene and surrounding genome. and flanking sequences. The genomic distances from 25 to parental genome. Hybridization
AND
’
HI -
ENGH 200 I
250 I
27310 kb 4
II-45 i
SE24
’ I
’ =
PA36 i
as determined by restriction enzyme mapping (7, 8) of eight cosmids that contain the Chinese hamster DHFR These fragments were cloned from a methotrexate-resistant CHO line with multiple copies of the DHFR gene 273 kb shown represent the repeated unit, or amplicon. Selected pairs of cosmid probes provided a range of 220 kb. pA36 contains the junction between amplicons and, therefore, sequences separated by 250 kb in the of single cosmids served as 0-kb controls.
clear boundaries can be seen upon fluorescein excitation. The efficiency of target labeling in interphase nuclei with FISH is high and the frequency of nonspecifically labeled sites is usually low. The frequency of UA41 nuclei labeled with each of the eight cosmid probes alone varied from 89 to 100% with a mean of 97% in both experiments described below (Table 1). Of 242 nuclei labeled simultaneously with all eight cosmids, only 1 showed a fluorescent spot at a second site in these hemizygote nuclei. Thus, in most experiment,s, >99.5% of the bright fluorescent spots within the nuclear boundary represent true sites of hybridization. To determine the capability of FISH to resolve closely spaced sequences in interphase nuclei, eight pairs of DHFR-region cosmids and probe pA36 alone were selected to provide a range of distances between probe midpoints of 25-250 kb. The two cosmids in each pair were hybridized simultaneously to UA41 interphase nuclei. Color photographs were made of a random selection of at least 20 hybridized interphase nuclei for each cosmid pair using a Zeiss Axiophot microscope. Photographic images were projected on a wall, and the distance between the centers of fluorescent spots was determined. Nuclei were scored in a coded fashion to prevent observer bias. UA41 were grown in monolayer, and their nuclei were relatively flat. When affixed to slides, the nuclei were usually sufficiently flat so that both hybridization dots were in focus during photography. When this was not the case, the nucleus was eliminated from distance calculations. A single, round hybridization spot in a nucleus was scored a distance of 0. Often, the fluorescent site was oblong rather than round, because two fluorescent spots were not completely resolved. In these cases, the distance between the apparent centers of the spots making up the com-
posite site was determined. Because of the high efficiency of target labeling observed with single cosmids on parallel slides (Table l), we conclude that unresolved or merged spots are the result of chromatin orientation or compaction in the target nucleus, rather than the failure to label one of the two probe sites. A hemacytometer grid photographed at the same magnification and projected similarly was used to convert measured, projected distances to actual distances (1 cm = 0.7 wm). With increasing genomic distance between cosmid inserts, the frequency with which the hybridization site could be resolved into separate spots and the mean distance between the spots increased (Fig. 3). For example, the spot separation for probes Q23 and Hl, whose inserts are separated by 90 kb, ranged from 0 to 1.4 pm with a mean of 0.45 pm. Representative UA41 nuclei probed with cosmid pairs whose inserts are separated by 90 and 195 kb are shown in Figs. 2C and 2D, respectively. A second experiment was performed (i) to assess the reproducibility of the technique and (ii) to determine if the genomic order of the probes could be derived from relative interphase distances. For this, microscope slides of interphase cells were prepared from freshly thawed UA41 cells after 4 days at confluency. Cosmids were hybridized singly to verify high hybridization efficiency as reported above. Hybridization was carried out, with 18 different pairs of probes, as well as with probe pA36 alone. Four probe pairs were done in duplicate to test the effect of the buffer in which the avidin is applied on measured interphase distance. Probe names were coded, microscope slides were coded during photography, and photographic slides were randomized during projection and measurement.. In this experiment, slides for three probe pairs were rejected because either the signal was too dim for photography or non-
FIG. 2. Fluorescence hybridization detection and localization of DHFR-region cosmid probes to metaphase and interphase nuclei. (A) Metaphase cell of the hemizygote UA41 hybridized with two cosmids (32Al and SE24) showing the single copy of the DHFR genomic region in this line. (B) UA41 interphase nuclei hybridized with a single cosmid, N27, showing a single, tightly localized fluorescent spot in each of the three nuclei. The more intensely fluorescent, amorphous areas within the nuclei are probably nucleoli. (C) Three UA41 nuclei hybridized with N27 and 1145, two cosmids with with two cosmids (Q23 and Hl) whose genomic separation is =90 kb. (D) Two UA41 nuclei hybridized known genomic separation of z195 kb. (E) Metaphase cell of DG46, an inversion mutant of UA41, hybridized with 32Al and SE24. The DHFR-region cosmids are separated by roughly 50 Mb of DNA. (DG46 also has a q-arm duplication in 22, giving this chromosome a DNA content of x480 Mb). The large, diffusely intense region in lower chromosome was red in original color image and does not represent a true hybridization site. In this black and white image, it appears white. (F) DG46 interphase nuclei hybridized with two cosmids flanking the inversion breakpoint (32Al and SE24).
714
TRASK,
Efficiency
of Target
TABLE
1
Labeling
Using % Nuclei
Cosmid N27
Q23 32Al 26A31 HI SE24 II45 pA36
Expt
1
100 99 97 n.d. 90 97 98 96
Note. Expt 1 (Fig. 3): ~100 nuclei 2 (Fig. 4): 40-70 nuclei counted.
PINKEL,
Singie
AND
Cosmids
labeled Expt
2
100 97 100 100 89 98 98 100 counted
for each cosmid;
Expt
specific avidin binding was excessive. The mean interphase distance was determined as described above for the remaining probe pairs. After probe decoding, the relationship between mean distance between hybridization sites in interphase and known genomic distance was determined and is shown in Fig. 4. For example, the interphase distante measured between probes 1145and 26A31, which are separated in the genome by 112 kb, ranged from 0 to 0.8 pm, with a mean of 0.37 pm. The measured distance between the two portions of pA36, separated in the genome by 250 kb, ranged from 0 to 1.7 pm, with a mean of 0.69 grn. The relationship is similar in both relative and absolute terms to that observed in the smaller first experiment. Measured values do not increase monotonically. However, genomic distances predicted from measured distances using a straight line through the points as a standard curve deviate ~40 kb from expected genomic distances. No significant difference in measured interphase distances was seen between the trials using different buffers for avidin application. The most likely order of seven probes was derived from relative interphase distances without knowledge of the nature of the relationship between interphase and genomic distance (Fig. 5). This is possible because nested sets, or three pairwise combinations of three probes, provide order information. For example, the interphase distance between N27 and SE24 was greater than those between either N27 and Hl or Hl and SE24. Thus, the most likely order is N27-Hl-SE24. Ambiguities in the order of N27-Q23 and of 32Al-26A31 were resolved by the presence of nested sets and pair combinations that overlapped the samegenomic region. For example, although distances for one nested set, N27-Q23-32A1, would predict the order Q23-N2732A1, distances for two other nested sets, N27-Q23SE24 and N27-Q23-26A31, predict the order to be N27-Q23-etc. The latter order is further corroborated
VAN
DEN
ENGH
by the observation that the distance N27-Hl is greater than the distance &23-Hl. The order determined in this way from interphase distances matched the order determined by restriction enzyme mapping. The availability of three different DHFR-deficient inversion mutants of UA4I (Urlaub et al., 1983, 1986) allowed investigation of the relationship between interphase distance and genomic distance for sequences separated by several chromosome bands. From cytogenetic analysis, the distance between two sequences that flank the inversion breakpoint (32A1, SE24) was estimated to be 50,60, and 90 Mb in DG46, DG24, and DG23, respectively. They are separated by 100 kb in UA41. Metaphase and interphase cells of DG46 probed with the two cosmids are shown in Figs. 2E and 2F. Figure 2E illustrates also that the distance between 100 A
0
100
200
3 0
100
200
300
0.0II 0 known
distance between midpoints of probes (kb)
of hybridization sites that appeared to FIG. 3. (A) Frequency be single spots after hybridization of selected pairs of DHFR region cosmids to UA41, a CHO line carrying only one copy of this region. (B) Relationship of mean interphase distance to genomic distance in Experiment 1. Error bars indicate standard error of mean. The numbers of nuclei analyzed for each genomic separation data point from 0 to 250 kb were 26, 33, 37. 30, 24, 22, 25, 20, 36, 36.
SEQUENCE
PROXIMITY
001
IN
1
0
100
known
distance
200
between of probes (kb)
3 O[
midpolnts
FIG. 4. Relationship of mean interphase distance to genomic distance in Experiment 2. Error bars indicate standard error of mean. More than 30 nuclei were analyzed for each data point. The data at 0 kb represent the results using each of the seven cosmids alone. Avidin-FITC was applied to slides in two alternative buffers: phosphate buffer/O.105 NP-40/5% dry milk or 4X SSC/l% BSA (solid diamonds and open squares, respectively).
hybridization sites on sister chromatids is not always equal. The mean interphase distances between probes in nuclei of the inversion mutants are shown in Table 2 and are an order of magnitude greater than those observed in UA41 nuclei. For the small number of points studied in this range, interphase distance increases with genomic distance. DISCUSSION
Our data indicate that, over the range from 25 to at least 250 kb, the distance between hybridization sites in interphase nuclei can be used to estimate the genomic distance between sequencesin the CHO DHFR region. Estimates of genomic distance to within 40 kb are possible by analyzing interphase distance in the small number of nuclei used in this study. Analyzing more nuclei will increase the reliability of mean interphase distance estimates to allow mapping or ordering to be done on a finer scale and chromatin packing between closely spaced sequencesto be studied. Despite the relative inaccuracy and the nonmonotonic nature of the measured relationship, interphase distance measurement,sof nested sets of probe pairs that overlap the same genomic region could be used to correctly derive the genomic order of the probes. The lower limit of the range of genomic distances that can be studied with FISH and interphase mapping is dictated by several factors: (i) the minimum distance at which two fluorescent dots can be completely resolved with a high quality fluorescence microscope (20.24 pm at 510 nm with 1.3 n.a. objective (Darnell
INTERPHASE
CEI,L
NUCLEI
715
et al., 1986)); (ii) the necessity for efficiently depositing a sufficiently large cluster of fluorescein molecules at the site of hybridization for visualization; and (iii) the degree of chromatin packing and its variation from cell to cell. At the other extreme, the slope of the relationship between interphase distance and genomic distance is relatively low in the 50-90 Mb range. At genomic distances >50 Mb, sequence mapping by FISH to metaphase or prometaphase chromosomes should yield equivalent or better estimates of genomic distances, supplemented with position data relative to chromosome landmarks. For mapping and ordering purposes, a straight line may be a reasonable approximation of the relationship between interphase and genomic distance within restricted ranges of genomic distance. The true relationship may not be monotonic even over short ranges of genomic distances. It may be different for different regions of the genome, depending on transcriptional activity or methylation. It is clear, however, that the slope of the relationship must decrease with increasing genomic distance, as chromatin is extensively packaged to fit within the nuclear membrane. Indeed, the slope of a straight line fit to the data in the 50-90 Mb range is 30 times lower than the slope of a line fit to the data in the O-250 kb range. Extrapolation of a linear relationship from the 50-90 Mb range results in interphase distances exceeding the average nuclear diameter of the cells studied (UA41 = 17 pm; inversion mutants z 21-25 pm) for sequences>200 Mb apart. At genomic distances >50 Mb, the relationship of interphase to genomic distance may also be affected by factors such as the proximity of sequences to the centromere and the arrangement and shape of chromosome domains in interphase. It is tempting to infer the nature of chromatin or nuclear organization from the data presented. Packing ratios can be calculated from the mean distance observed between sequences in interphase. Two points should be made in this regard: (i) target cells in this study and that of Lawrence et al. (1988) were subjected to hypotonic swelling, acid-ethanol fixation, flattening on slides, and salts and formamide in the hybridization buffer, and, therefore, measured distances need not reflect interphase proximity and chromatin organization in the untreated, living cell nucleus; and (ii) measured distances underestimate actual distances by a factor, x/4, as a result of the orthogonal projection of hybridization site positions onto the observation plane. A mean distance of 0.59 km between sequences 160 kb apart (observed in both experiments) implies a 70.fold overall shortening of the linear DNA molecule (at 0.34 nm/bp, 164 kb of a linear DNA molecule = 56 pm; 56 X 7r/4 = 44 pm). Packing orders of 90-100 can be similarly calculated for the observed distances between sequences 250 kb apart (0.76 and 0.69 pm (Experiments
716
TRASK,
PINKEL,
AND
VAN
DEN
ENGH
order derived from 3 probes in 3 paired combinations: N27
’
N27
I
N27
I
N27
I
I SE24 26A31
26A31
2eA3i
(Prim)
1 SE24 Hi
32Al (pnm)
32Al 023
Q23
N27 I
Hl
32Al
I SE24
N27 - 26A31 ;;;
SE24
] :;;,;;;24
f$’
N27.
Q23
(32Al
.26A31)
Q23 - 32Al
26A31
Hl - SE24
/ SE24
(prim)
1 SE24
N27
Q23 - SE24
Q23
023 I
Ill45 26A31
Q23 I 32A1 Cl23 1 Q23 I
(v-4
26A31
32Al I 32Al
Ill45
Q23
II145
023
26A31 - 32Al
- 1145
- Hl
SE24
1145
- 1145
Hi I (Prim)
N271
I26A31 023
N271
N27.
Q23 - 26A31
Q23
N27 - 26A31
J 26A31 Q23
(mm)
Q23 1
I32Al N27 I
Q23 I
32A’ I32Al
N27 0.2pm
predicted
order
-0 N27 - Q23 - 32Al
- 26A31 - HI
SE24 - 1145
FIG. 5. Relative interphase distances observed in Experiment 2 displayed to illustrate how the correct order of seven probes was derived. (The eighth probe, pA36, was excluded, since it contains sequences at both the beginning and end of this range, and was not paired with other probes in the experiment.) The lengths of lines between probe names are proportional to the mean measured interphase distance between them (also plotted in Fig. 4 after decoding). The derived order agrees with restriction enzyme mapping results shown in Fig. 1.
1 and 2, respectively) versus 85 pm X 7~/4 = 67 pm). For comparison, the packing ratio for the 30-nm-diameter fiber of packed nucleosomes is -40, and the predicted packing ratio for DNA in the looped domain conformation is 480 (Nelson et al., 1986). In metaphase chromosomes, roughly 1 m of DNA helix is compressed 12,000-fold into an overall length of roughly 85 pm. Thus, our measurements indicate that the DNA between sequences loo-250 kb apart is 2- to 3-fold more condensed on average in target interphase nuclei than it is in the 30-nm fiber conformation. The calculated packing ratio increases as the genomic distance between sequences increases, suggesting the presence of higher orders of folding between more distant sequences with the possible further constraints of nuclear or chromosomal domain boundaries. DHFR sequences separated in the inversion mutant, DG46, by 50 Mb or 17 mm of linear DNA helix had a mean separation in interphase of 3.6 pm, giving a packing ratio of 3700. This is approximately three times lower than in metaphase chromosomes (see also Fig. 2E vs Fig. 2F). If the observations made with this set of DHFRregion cosmids can be extended to other probes in other
regions of the genome, interphase chromatin mapping should be of use in genome ordering efforts. Techniques employing somatic cell hybrids containing chromosome fragments, in situ hybridization to metaphase spreads, or recombination frequency in pedigrees are limited in their application to genomic distances greater than 1 Mb. Pulsed field gel electrophoresis and yeast artificial chromosome cloning provide genomic distance information for sequences
SEQUENCE
TABLE Interphase
Distance Breakpoints
Cell line UA41 DG46 DG24 DG23
0.10 53 60 90 used: 32Al
IN
2
Number of nuclei 22 20 17 30
INTERPHASE
3.
in Cell Lines with Inversion in the DHFR Locus
Known distance between probe midpoints (Mb)”
a Probes
PROXIMITY
Measured distance between probes in interphase mean + SEM (Frnl 0.23 3.6 4.8 7.8
_C 0.06 + 0.6 + 0.6 t 1.0
and SE24.
the relative order of sequences in the genome. Interphase distance measurements may also be a means for detecting deletions of 100-1000 kb, a range that links the detection capabilities of electrophoretic and cytogenetic methods. ACKNOWLEDGMENTS The authors are grateful to Drs. J. Hamlin for providing cosmid clone preparations, L. Chasin and R. Schimke for providing cell lines, J. Gray and E. Branscomb for stimulating discussions and critical reading of the manuscript, and Ms. C. Lazes for cell culture and slide preparation assistance. Work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under Contract W-7405-ENG-48 with support from USPHS Grants HD17655 and CA45919.
4.
2.
717
FUNANAGE, V. L., AND MYODA, T. T. (1986). Localization of Chinese hamster dihydrofolate reductase gene to band p23 of chromosome 2. Somat. Cell Mol. Genet. 12: 6499655. JOHNSON, G. D., AND NOGUERIA, J. G. M. (1981). A simple method of reducing the fading of immunofluorescence during microscopy. J. Immunol. Methods 43: 349-350. LANDEGENT, J. E., JANSEN IN DE WAL, N., DIRKS, R. W., BAAS, F., AND VAN DER PLOEG, M. (1987). Use of whole cosmid cloned genomic sequences for chromosomal localization by non-radioactive in situ hybridization. Hum. Genet. 77: 366-370.
6.
LAWRENCE, J. B., VILLNAVE, C. A., AND SINGER, R. H. (1988). Sensitive, high-resolution chromatin and chromosome mapping in situ: Presence and orientation of two closely integrated copies of EBV in a lymphoma line. Cell 42: 51-61. LOONEY, J. E., AND HAMLIN, J. L. (1987). Isolation of the amplified dihydrofolate reductase domain from methotrexate-resistant Chinese hamster ovary cells. Mol. Cell Biol. 7: 569-577.
7.
8.
9.
10.
MA, C., LOONEY, J. E., LEII, T-H., AND HAMLIN, J. L. (1988). Organization and genesis of dihydrofolate reductase amplicons in the genome of a methotrexate-resistant Chinese hamster ovary cell line. Mol. Cell Biol. 8: 2316-2327. MANIATIS, T., FRITSCH, E. F., AND SAMBROOK, J. (1982). “Molecular Cloning: A Laboratory Manual,” Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. NELSON, W. G., PIENTAL, K. J., BARRACK, E. R., AND COFFEY, D. S. (1986). The role of the nuclear matrix in the organization and function of DNA. Annu. Reu. Biophys. C’hem. 15: 457475.
11.
PINKEL, D., LANDEGENT, J., COLLINS, C., FUSCOE, J., SEGRAVES, R., LUCAS, J., AND GRAY, J. W. (1988). Fluorescence in situ hybridization with human chromosome-specific libraries: Detection of trisomy 21 and translocations of chromosome 4. Proc. Natl. Acad. Sci. USA 85: 9138-9142.
12.
PINKEL, D., STRAUME, T., AND GRAY, J. W. (1986). Cytogenetic analysis using quantitative, high-sensitivity, fluorescence hybridization. Proc. Notl. Acad. Sci. USA 83: 2934-2938. URLAUB, G., KAS, E., CAROTHERS, A. M., AND CHASIN, L. A. (1983). Deletion of the diploid dihydrofolate reductase locus from cultured mammalian cells. Cell 33: 405-412.
13. DARNELL, J., LODISH, H., AND BALTIMORE, D. (1986). “Molecular Cell Biology,” Sci. Amer. Books, New York. DEAVEN, L. L., AND PETERSON, D. F. (1973). The chromosomes of CHO, an aneuploid Chinese hamster cell line: G-band, Cband, and autoradiographic analyses. Chromosoma 41: 129144.
NUCLEI
5.
REFERENCES 1.
CELL
14.
URLAUB, G., MITCHELL, P. J., KAS, E., CHASIN, L. A., FLINANAGE, V. L., MYODA, T. T., AND HAMLIN, J. (1986). Effect of gamma rays at the dihydrofolate reductase locus: Deletions and inversions. Somat. Cell Mol. Genet. 12: 5555566.
5, 710-717
(1989)
The Proximity of DNA Sequences in Interphase Cell Nuclei Is Correlated to Genomic Distance and Permits Ordering of Cosmids Spanning 250 Kilobase Pairs’ BARBARA TRASK,’ Biomedical
Sciences Divkon,
Lawrence
DAN PINKEL, AND GER VAN DEN ENGH
LIvermore National Laboratory, Received
April
5, 1989,
revised
Inc.
INTRODUCTION The concept of interphase chromatin mapping to determine the relative distance between two sequences in the genome was recently introduced by Lawrence et al. (1988). The technique relies on specific labeling of single copy target sequences in interphase nuclei with ’ The U.S. Government’s right to retain a nonexclusive free license in and to the copyright covering this paper, mental purposes, is acknowledged. * To whom correspondence should he addressed.
0888.7543/89
Copyright All rights
$3.00
8 1989 by Academic Press. Inc. of reproduction in any form reserved.
15.
California
94550
1989
fluorescence in situ hybridization (FISH). With FISH, complementary DNA sequence probes are hybridized to nuclei and are subsequently detected with a fluorescent label, resulting in a sharply defined fluorescent spot at the site of hybridization. Lawrence et al. (1988) determined the orientation of two integrated viral genomes from the intensity and spacing of the fluorescent spots obtained by hybridizing DNA sequence probes of different portions of the EpsteinBarr virus t.o nuclei from infected cells. They assumed the relationship bet.ween genomic distance and interphase distance to be linear with zero-intercept to conclude that the viral sequences were separated by 340 kb and to estimate a packing order of 100 for interphase chromatin. This work led to the hope that. analysis of probe separation in interphase could be used to map more closely spaced sequences and could be developed into a tool for genome ordering studies. For this application, the relationship between interphase distance (in pm) and genomic distance (in kb) over a range of genomic distances must be determined. A set of cosmids containing fragments of the Chinese hamst,er (CHO) genome including and surrounding the dihydrofolate reductase (DHFR) gene has been isolat,ed and characterized by Hamlin and co-workers (Looney and Hamlin, 1987; Ma et al., 1988).. This set of cosmids provides an opportunit,y to study the relationship of physical distance between hybridization sites in interphase nuclei to known genomic distance between fragments over a range of 25 to 250 kb. Cells carrying various chromosome inversions that disrupt the DHFR gene were used to further analyze interphase distance between sequences separated by 50-90 Mb.
The physical distance between DNA sequences in interphase nuclei was determined using eight cosmids containing fragments of the Chinese hamster genome that span 273 kb surrounding the dihydrofolate reductase (DHFR) gene. The distance between these sequences at the molecular level has been determined previously by restriction enzyme mapping (J. E. Looney and J. L. Hamlin, 1987, Mol. Cell Biol. 7: 569577; C. Ma et al., 1988, Mol. Cell Biol. 8: 2316-2327). Fluorescence in situ hybridization was used to localize the DNA sequences in interphase nuclei of cells bearing only one copy of this genomic region. The distance between DNA sequences in interphase nuclei was correlated to molecular distance over a range of 25 to at least 250 kb. The observed relationship was such that genomic distance could be predicted to within 40 kb from interphase distance. The correct order of seven probes was derived from interphase distances measured for 19 pair-wise combinations of the probes. Measured distances between sequences = 200 kb apart indicate that the DNA is condensed 70- to loo-fold in hybridized nuclei relative to a linear DNA helix molecule. Cell lines with chromosome inversions were used to show that interphase distance increases with genomic distance in the 50-90 Mb range, but less 8~ 1989 Academic steeply than in the 25-250 kb range. Press.
June
P.O. Box 5507 L-452, Livermore,
EXPERIMENTAL
METHODS
Cell Lines
royaltyfor govern-
1JA41 is a y-ray mutant in which the DHFR gene on chromosome 2 has been deleted leaving only one copy on chromosome 22 (Urlaub et al., 1983). DG46, 710
SEQUENCE
PROXIMITY
IN
DG24, and DG23 are DHFR-deficient mutants, obtained after y-ray treatment of UA41, in which the DHFR locus is disrupted by inversion in metaphase chromosomes. The inversions in DG46 (inv(Z2) dup(Z2) (pter -+ ~23: :p14 -W p23::p14 -+ q37: :q24 q37)) and DG24 (inv(Z2)(p23p31)) were detected by banding (Urlaub et al., 1986) and by FISH to metaphase chromosomes (B. J. Trask, unpublished observations); the inversion in DG23 (inv(Z2) (~23~12)) was detected by FISH to metaphase chromosomes (ibid.). The distance in metaphase between breakpoints in the inversion mutants was measured as a fraction of the length of the 22 chromosome. The DNA content of the inverted segment then was calculated using an estimate oft he DNA content of 22 (390 Mb), based on the DNA content of CHO chromosome 2 (Funanage and Myoda, 1986) and the relative idiogram lengths of chromosomes 22 and 2 (Deaven and Peterson, 1973). Slide Preparation Interphase cells were harvested by trypsinization after 2-4 days growth at complete confluency, incubated 15 min in 75 mMKC1 at 37”C, fixed in methanol:acetic acid (3:1), and dropped on slides. Several observations led us to consider these preparations to contain a relatively pure population of cells in G, : (i) the size of all nuclei was approximately the same, suggesting a low frequency of cells in G,; (ii) no mitotic cells were observed on these slides; and (iii) few cells were observed to be labeled with two fluorescent spots when single probes were hybridized (see Results). Metaphase spreads were prepared from cells collected by mitotic shake-off from nonconfluent cultures incubated 1.5 h in medium containing 0.1 pg/ml colcemid. Slides were baked 4 h at 65°C to artificially age them and stored at -20°C in N:!-enriched atmosphere until use. Probes and Hybridization Purified cosmid DNA was biotinylated by nick translation with biotin-ll-dUTP (Bethesda Research Laboratories). Probe modification averaged 35%. Probes were hybridized to cells using a modification of a published procedure (Pinkel et al.. 1986). Briefly, slides were treated with RNase (100 pg/ml in 2X SSC) prior to being denatured for 2 min at 70°C in 70% formamide/2x SSC and then quenched in cold 70% ethanol. After dehydration, 10 ~1 of a hybridization mixture containing 1-2 ng/pl probe or pooled probes, 50% formamide, 2X SSC, 10% dextran sulfate, and 1 mg/ml Chinese hamster genomic DNA (purified (Maniatis et al., 1982) from AA8 cells, sonicated to 200-600 bp, and resuspended after ethanol precipitation at 10 mg/ml in 1X TE) was applied to slides. Slides were incubated overnight and then washed in 50% formamide/2X SSC and in 2~ SSC. Avidin conjugated
INTERPHASE
CELL
711
NIICLEI
to fluorescein (DCS, Vector Laboratories) was applied to the slides for 30 min at 37°C in PNM buffer (0.1 M NaH2P04, 0.1 M Na2HP04, 0.1% NP-40, 5% nonfat dry milk, pH 8) or in 4X SSC containing 1% BSA (Lawrence et al., 1988). Free avidin was removed by washing slides in several changes of PNM buffer without milk or in 4~ SSC/l% Triton X-100, depending on the buffer used for avidin incubation. Distance
Analysis
Cells were counterstained and viewed in DAPI present at 1 pg/ml in a fluorescence antifade solution (Johnson and Nogueria, 1981). Color photographs (Kodak Ektachrome 400 t,ransparency film, 90 s exposure) were made of a random selection of at least 20 hybridized interphase nuclei for each cosmid pair using a Zeiss Axiophot microscope equipped with a Zeiss Plan Neofluor oil immersion objective (100X, 1.3 n.a.) and 1.25X optivar (epifluorescence filters for fluorescein: excitation BP485 nm, reflector 510 nm, emission LP 520 nm). Photographic images were projected on a wall, and the distance between the centers of fluorescent spots was determined as described in text. Figures are black and white renditions of color images. RESULTS
The relative position of the eight genomic fragments used in this study has been determined by restriction enzyme mapping (Looney and Hamlin, 1987; Ma et al., 1988) and is shown in Fig. 1. The fragments were cloned from a methotrexate-resistant CHO line with multiple copies of the DHFR gene and flanking sequences. The 273 kb shown in Fig. 1 represent the repeated unit, or amplicon. One cosmid, pA36, contains the junction between amplicons and, therefore, sequences separated by 250 kb in the parental genome (Ma et al., 1988). Cosmids were biotinylated by nick translation and hybridized singly or in pairs to metaphase and interphase cells fixed on slides. Hybridization of repetitive sequences in cosmid probes to target cells was blocked by the addition of sonicated CHO genomic DNA to the hybridization solution (Landegent et al., 1987). The hybridization sites were visualized by a single incubation of slides in fluorescein-conjugated avidin. UA41, a CHO cell line with only one copy of the DHFR region sequences (Urlaub et al., 1983), was chosen as hybridization target to simplify analysis of hybridization sites. Hybridization of two probes, 32Al and SE24, to the metaphase cell in Fig. 2A demonstrates the single chromosomal site for these sequences in UA41. Hybridization of a single cosmid to UA41 interphase nuclei (Fig. 2s) illustrates the small (0.30.6 pm diameter) fluorescent hybridization site obtained with FISH. A low amount of hybridization of repetitive elements in the probe is useful because nu-
712
TRASK, 0
50
N27
PINKEL,
VAN
DEN
150
DHFR+
100
32Al
I
‘26A31 t
FIG. 1. Relative spacing gene and surrounding genome. and flanking sequences. The genomic distances from 25 to parental genome. Hybridization
AND
’
HI -
ENGH 200 I
250 I
27310 kb 4
II-45 i
SE24
’ I
’ =
PA36 i
as determined by restriction enzyme mapping (7, 8) of eight cosmids that contain the Chinese hamster DHFR These fragments were cloned from a methotrexate-resistant CHO line with multiple copies of the DHFR gene 273 kb shown represent the repeated unit, or amplicon. Selected pairs of cosmid probes provided a range of 220 kb. pA36 contains the junction between amplicons and, therefore, sequences separated by 250 kb in the of single cosmids served as 0-kb controls.
clear boundaries can be seen upon fluorescein excitation. The efficiency of target labeling in interphase nuclei with FISH is high and the frequency of nonspecifically labeled sites is usually low. The frequency of UA41 nuclei labeled with each of the eight cosmid probes alone varied from 89 to 100% with a mean of 97% in both experiments described below (Table 1). Of 242 nuclei labeled simultaneously with all eight cosmids, only 1 showed a fluorescent spot at a second site in these hemizygote nuclei. Thus, in most experiment,s, >99.5% of the bright fluorescent spots within the nuclear boundary represent true sites of hybridization. To determine the capability of FISH to resolve closely spaced sequences in interphase nuclei, eight pairs of DHFR-region cosmids and probe pA36 alone were selected to provide a range of distances between probe midpoints of 25-250 kb. The two cosmids in each pair were hybridized simultaneously to UA41 interphase nuclei. Color photographs were made of a random selection of at least 20 hybridized interphase nuclei for each cosmid pair using a Zeiss Axiophot microscope. Photographic images were projected on a wall, and the distance between the centers of fluorescent spots was determined. Nuclei were scored in a coded fashion to prevent observer bias. UA41 were grown in monolayer, and their nuclei were relatively flat. When affixed to slides, the nuclei were usually sufficiently flat so that both hybridization dots were in focus during photography. When this was not the case, the nucleus was eliminated from distance calculations. A single, round hybridization spot in a nucleus was scored a distance of 0. Often, the fluorescent site was oblong rather than round, because two fluorescent spots were not completely resolved. In these cases, the distance between the apparent centers of the spots making up the com-
posite site was determined. Because of the high efficiency of target labeling observed with single cosmids on parallel slides (Table l), we conclude that unresolved or merged spots are the result of chromatin orientation or compaction in the target nucleus, rather than the failure to label one of the two probe sites. A hemacytometer grid photographed at the same magnification and projected similarly was used to convert measured, projected distances to actual distances (1 cm = 0.7 wm). With increasing genomic distance between cosmid inserts, the frequency with which the hybridization site could be resolved into separate spots and the mean distance between the spots increased (Fig. 3). For example, the spot separation for probes Q23 and Hl, whose inserts are separated by 90 kb, ranged from 0 to 1.4 pm with a mean of 0.45 pm. Representative UA41 nuclei probed with cosmid pairs whose inserts are separated by 90 and 195 kb are shown in Figs. 2C and 2D, respectively. A second experiment was performed (i) to assess the reproducibility of the technique and (ii) to determine if the genomic order of the probes could be derived from relative interphase distances. For this, microscope slides of interphase cells were prepared from freshly thawed UA41 cells after 4 days at confluency. Cosmids were hybridized singly to verify high hybridization efficiency as reported above. Hybridization was carried out, with 18 different pairs of probes, as well as with probe pA36 alone. Four probe pairs were done in duplicate to test the effect of the buffer in which the avidin is applied on measured interphase distance. Probe names were coded, microscope slides were coded during photography, and photographic slides were randomized during projection and measurement.. In this experiment, slides for three probe pairs were rejected because either the signal was too dim for photography or non-
FIG. 2. Fluorescence hybridization detection and localization of DHFR-region cosmid probes to metaphase and interphase nuclei. (A) Metaphase cell of the hemizygote UA41 hybridized with two cosmids (32Al and SE24) showing the single copy of the DHFR genomic region in this line. (B) UA41 interphase nuclei hybridized with a single cosmid, N27, showing a single, tightly localized fluorescent spot in each of the three nuclei. The more intensely fluorescent, amorphous areas within the nuclei are probably nucleoli. (C) Three UA41 nuclei hybridized with N27 and 1145, two cosmids with with two cosmids (Q23 and Hl) whose genomic separation is =90 kb. (D) Two UA41 nuclei hybridized known genomic separation of z195 kb. (E) Metaphase cell of DG46, an inversion mutant of UA41, hybridized with 32Al and SE24. The DHFR-region cosmids are separated by roughly 50 Mb of DNA. (DG46 also has a q-arm duplication in 22, giving this chromosome a DNA content of x480 Mb). The large, diffusely intense region in lower chromosome was red in original color image and does not represent a true hybridization site. In this black and white image, it appears white. (F) DG46 interphase nuclei hybridized with two cosmids flanking the inversion breakpoint (32Al and SE24).
714
TRASK,
Efficiency
of Target
TABLE
1
Labeling
Using % Nuclei
Cosmid N27
Q23 32Al 26A31 HI SE24 II45 pA36
Expt
1
100 99 97 n.d. 90 97 98 96
Note. Expt 1 (Fig. 3): ~100 nuclei 2 (Fig. 4): 40-70 nuclei counted.
PINKEL,
Singie
AND
Cosmids
labeled Expt
2
100 97 100 100 89 98 98 100 counted
for each cosmid;
Expt
specific avidin binding was excessive. The mean interphase distance was determined as described above for the remaining probe pairs. After probe decoding, the relationship between mean distance between hybridization sites in interphase and known genomic distance was determined and is shown in Fig. 4. For example, the interphase distante measured between probes 1145and 26A31, which are separated in the genome by 112 kb, ranged from 0 to 0.8 pm, with a mean of 0.37 pm. The measured distance between the two portions of pA36, separated in the genome by 250 kb, ranged from 0 to 1.7 pm, with a mean of 0.69 grn. The relationship is similar in both relative and absolute terms to that observed in the smaller first experiment. Measured values do not increase monotonically. However, genomic distances predicted from measured distances using a straight line through the points as a standard curve deviate ~40 kb from expected genomic distances. No significant difference in measured interphase distances was seen between the trials using different buffers for avidin application. The most likely order of seven probes was derived from relative interphase distances without knowledge of the nature of the relationship between interphase and genomic distance (Fig. 5). This is possible because nested sets, or three pairwise combinations of three probes, provide order information. For example, the interphase distance between N27 and SE24 was greater than those between either N27 and Hl or Hl and SE24. Thus, the most likely order is N27-Hl-SE24. Ambiguities in the order of N27-Q23 and of 32Al-26A31 were resolved by the presence of nested sets and pair combinations that overlapped the samegenomic region. For example, although distances for one nested set, N27-Q23-32A1, would predict the order Q23-N2732A1, distances for two other nested sets, N27-Q23SE24 and N27-Q23-26A31, predict the order to be N27-Q23-etc. The latter order is further corroborated
VAN
DEN
ENGH
by the observation that the distance N27-Hl is greater than the distance &23-Hl. The order determined in this way from interphase distances matched the order determined by restriction enzyme mapping. The availability of three different DHFR-deficient inversion mutants of UA4I (Urlaub et al., 1983, 1986) allowed investigation of the relationship between interphase distance and genomic distance for sequences separated by several chromosome bands. From cytogenetic analysis, the distance between two sequences that flank the inversion breakpoint (32A1, SE24) was estimated to be 50,60, and 90 Mb in DG46, DG24, and DG23, respectively. They are separated by 100 kb in UA41. Metaphase and interphase cells of DG46 probed with the two cosmids are shown in Figs. 2E and 2F. Figure 2E illustrates also that the distance between 100 A
0
100
200
3 0
100
200
300
0.0II 0 known
distance between midpoints of probes (kb)
of hybridization sites that appeared to FIG. 3. (A) Frequency be single spots after hybridization of selected pairs of DHFR region cosmids to UA41, a CHO line carrying only one copy of this region. (B) Relationship of mean interphase distance to genomic distance in Experiment 1. Error bars indicate standard error of mean. The numbers of nuclei analyzed for each genomic separation data point from 0 to 250 kb were 26, 33, 37. 30, 24, 22, 25, 20, 36, 36.
SEQUENCE
PROXIMITY
001
IN
1
0
100
known
distance
200
between of probes (kb)
3 O[
midpolnts
FIG. 4. Relationship of mean interphase distance to genomic distance in Experiment 2. Error bars indicate standard error of mean. More than 30 nuclei were analyzed for each data point. The data at 0 kb represent the results using each of the seven cosmids alone. Avidin-FITC was applied to slides in two alternative buffers: phosphate buffer/O.105 NP-40/5% dry milk or 4X SSC/l% BSA (solid diamonds and open squares, respectively).
hybridization sites on sister chromatids is not always equal. The mean interphase distances between probes in nuclei of the inversion mutants are shown in Table 2 and are an order of magnitude greater than those observed in UA41 nuclei. For the small number of points studied in this range, interphase distance increases with genomic distance. DISCUSSION
Our data indicate that, over the range from 25 to at least 250 kb, the distance between hybridization sites in interphase nuclei can be used to estimate the genomic distance between sequencesin the CHO DHFR region. Estimates of genomic distance to within 40 kb are possible by analyzing interphase distance in the small number of nuclei used in this study. Analyzing more nuclei will increase the reliability of mean interphase distance estimates to allow mapping or ordering to be done on a finer scale and chromatin packing between closely spaced sequencesto be studied. Despite the relative inaccuracy and the nonmonotonic nature of the measured relationship, interphase distance measurement,sof nested sets of probe pairs that overlap the same genomic region could be used to correctly derive the genomic order of the probes. The lower limit of the range of genomic distances that can be studied with FISH and interphase mapping is dictated by several factors: (i) the minimum distance at which two fluorescent dots can be completely resolved with a high quality fluorescence microscope (20.24 pm at 510 nm with 1.3 n.a. objective (Darnell
INTERPHASE
CEI,L
NUCLEI
715
et al., 1986)); (ii) the necessity for efficiently depositing a sufficiently large cluster of fluorescein molecules at the site of hybridization for visualization; and (iii) the degree of chromatin packing and its variation from cell to cell. At the other extreme, the slope of the relationship between interphase distance and genomic distance is relatively low in the 50-90 Mb range. At genomic distances >50 Mb, sequence mapping by FISH to metaphase or prometaphase chromosomes should yield equivalent or better estimates of genomic distances, supplemented with position data relative to chromosome landmarks. For mapping and ordering purposes, a straight line may be a reasonable approximation of the relationship between interphase and genomic distance within restricted ranges of genomic distance. The true relationship may not be monotonic even over short ranges of genomic distances. It may be different for different regions of the genome, depending on transcriptional activity or methylation. It is clear, however, that the slope of the relationship must decrease with increasing genomic distance, as chromatin is extensively packaged to fit within the nuclear membrane. Indeed, the slope of a straight line fit to the data in the 50-90 Mb range is 30 times lower than the slope of a line fit to the data in the O-250 kb range. Extrapolation of a linear relationship from the 50-90 Mb range results in interphase distances exceeding the average nuclear diameter of the cells studied (UA41 = 17 pm; inversion mutants z 21-25 pm) for sequences>200 Mb apart. At genomic distances >50 Mb, the relationship of interphase to genomic distance may also be affected by factors such as the proximity of sequences to the centromere and the arrangement and shape of chromosome domains in interphase. It is tempting to infer the nature of chromatin or nuclear organization from the data presented. Packing ratios can be calculated from the mean distance observed between sequences in interphase. Two points should be made in this regard: (i) target cells in this study and that of Lawrence et al. (1988) were subjected to hypotonic swelling, acid-ethanol fixation, flattening on slides, and salts and formamide in the hybridization buffer, and, therefore, measured distances need not reflect interphase proximity and chromatin organization in the untreated, living cell nucleus; and (ii) measured distances underestimate actual distances by a factor, x/4, as a result of the orthogonal projection of hybridization site positions onto the observation plane. A mean distance of 0.59 km between sequences 160 kb apart (observed in both experiments) implies a 70.fold overall shortening of the linear DNA molecule (at 0.34 nm/bp, 164 kb of a linear DNA molecule = 56 pm; 56 X 7r/4 = 44 pm). Packing orders of 90-100 can be similarly calculated for the observed distances between sequences 250 kb apart (0.76 and 0.69 pm (Experiments
716
TRASK,
PINKEL,
AND
VAN
DEN
ENGH
order derived from 3 probes in 3 paired combinations: N27
’
N27
I
N27
I
N27
I
I SE24 26A31
26A31
2eA3i
(Prim)
1 SE24 Hi
32Al (pnm)
32Al 023
Q23
N27 I
Hl
32Al
I SE24
N27 - 26A31 ;;;
SE24
] :;;,;;;24
f$’
N27.
Q23
(32Al
.26A31)
Q23 - 32Al
26A31
Hl - SE24
/ SE24
(prim)
1 SE24
N27
Q23 - SE24
Q23
023 I
Ill45 26A31
Q23 I 32A1 Cl23 1 Q23 I
(v-4
26A31
32Al I 32Al
Ill45
Q23
II145
023
26A31 - 32Al
- 1145
- Hl
SE24
1145
- 1145
Hi I (Prim)
N271
I26A31 023
N271
N27.
Q23 - 26A31
Q23
N27 - 26A31
J 26A31 Q23
(mm)
Q23 1
I32Al N27 I
Q23 I
32A’ I32Al
N27 0.2pm
predicted
order
-0 N27 - Q23 - 32Al
- 26A31 - HI
SE24 - 1145
FIG. 5. Relative interphase distances observed in Experiment 2 displayed to illustrate how the correct order of seven probes was derived. (The eighth probe, pA36, was excluded, since it contains sequences at both the beginning and end of this range, and was not paired with other probes in the experiment.) The lengths of lines between probe names are proportional to the mean measured interphase distance between them (also plotted in Fig. 4 after decoding). The derived order agrees with restriction enzyme mapping results shown in Fig. 1.
1 and 2, respectively) versus 85 pm X 7~/4 = 67 pm). For comparison, the packing ratio for the 30-nm-diameter fiber of packed nucleosomes is -40, and the predicted packing ratio for DNA in the looped domain conformation is 480 (Nelson et al., 1986). In metaphase chromosomes, roughly 1 m of DNA helix is compressed 12,000-fold into an overall length of roughly 85 pm. Thus, our measurements indicate that the DNA between sequences loo-250 kb apart is 2- to 3-fold more condensed on average in target interphase nuclei than it is in the 30-nm fiber conformation. The calculated packing ratio increases as the genomic distance between sequences increases, suggesting the presence of higher orders of folding between more distant sequences with the possible further constraints of nuclear or chromosomal domain boundaries. DHFR sequences separated in the inversion mutant, DG46, by 50 Mb or 17 mm of linear DNA helix had a mean separation in interphase of 3.6 pm, giving a packing ratio of 3700. This is approximately three times lower than in metaphase chromosomes (see also Fig. 2E vs Fig. 2F). If the observations made with this set of DHFRregion cosmids can be extended to other probes in other
regions of the genome, interphase chromatin mapping should be of use in genome ordering efforts. Techniques employing somatic cell hybrids containing chromosome fragments, in situ hybridization to metaphase spreads, or recombination frequency in pedigrees are limited in their application to genomic distances greater than 1 Mb. Pulsed field gel electrophoresis and yeast artificial chromosome cloning provide genomic distance information for sequences
SEQUENCE
TABLE Interphase
Distance Breakpoints
Cell line UA41 DG46 DG24 DG23
0.10 53 60 90 used: 32Al
IN
2
Number of nuclei 22 20 17 30
INTERPHASE
3.
in Cell Lines with Inversion in the DHFR Locus
Known distance between probe midpoints (Mb)”
a Probes
PROXIMITY
Measured distance between probes in interphase mean + SEM (Frnl 0.23 3.6 4.8 7.8
_C 0.06 + 0.6 + 0.6 t 1.0
and SE24.
the relative order of sequences in the genome. Interphase distance measurements may also be a means for detecting deletions of 100-1000 kb, a range that links the detection capabilities of electrophoretic and cytogenetic methods. ACKNOWLEDGMENTS The authors are grateful to Drs. J. Hamlin for providing cosmid clone preparations, L. Chasin and R. Schimke for providing cell lines, J. Gray and E. Branscomb for stimulating discussions and critical reading of the manuscript, and Ms. C. Lazes for cell culture and slide preparation assistance. Work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under Contract W-7405-ENG-48 with support from USPHS Grants HD17655 and CA45919.
4.
2.
717
FUNANAGE, V. L., AND MYODA, T. T. (1986). Localization of Chinese hamster dihydrofolate reductase gene to band p23 of chromosome 2. Somat. Cell Mol. Genet. 12: 6499655. JOHNSON, G. D., AND NOGUERIA, J. G. M. (1981). A simple method of reducing the fading of immunofluorescence during microscopy. J. Immunol. Methods 43: 349-350. LANDEGENT, J. E., JANSEN IN DE WAL, N., DIRKS, R. W., BAAS, F., AND VAN DER PLOEG, M. (1987). Use of whole cosmid cloned genomic sequences for chromosomal localization by non-radioactive in situ hybridization. Hum. Genet. 77: 366-370.
6.
LAWRENCE, J. B., VILLNAVE, C. A., AND SINGER, R. H. (1988). Sensitive, high-resolution chromatin and chromosome mapping in situ: Presence and orientation of two closely integrated copies of EBV in a lymphoma line. Cell 42: 51-61. LOONEY, J. E., AND HAMLIN, J. L. (1987). Isolation of the amplified dihydrofolate reductase domain from methotrexate-resistant Chinese hamster ovary cells. Mol. Cell Biol. 7: 569-577.
7.
8.
9.
10.
MA, C., LOONEY, J. E., LEII, T-H., AND HAMLIN, J. L. (1988). Organization and genesis of dihydrofolate reductase amplicons in the genome of a methotrexate-resistant Chinese hamster ovary cell line. Mol. Cell Biol. 8: 2316-2327. MANIATIS, T., FRITSCH, E. F., AND SAMBROOK, J. (1982). “Molecular Cloning: A Laboratory Manual,” Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. NELSON, W. G., PIENTAL, K. J., BARRACK, E. R., AND COFFEY, D. S. (1986). The role of the nuclear matrix in the organization and function of DNA. Annu. Reu. Biophys. C’hem. 15: 457475.
11.
PINKEL, D., LANDEGENT, J., COLLINS, C., FUSCOE, J., SEGRAVES, R., LUCAS, J., AND GRAY, J. W. (1988). Fluorescence in situ hybridization with human chromosome-specific libraries: Detection of trisomy 21 and translocations of chromosome 4. Proc. Natl. Acad. Sci. USA 85: 9138-9142.
12.
PINKEL, D., STRAUME, T., AND GRAY, J. W. (1986). Cytogenetic analysis using quantitative, high-sensitivity, fluorescence hybridization. Proc. Notl. Acad. Sci. USA 83: 2934-2938. URLAUB, G., KAS, E., CAROTHERS, A. M., AND CHASIN, L. A. (1983). Deletion of the diploid dihydrofolate reductase locus from cultured mammalian cells. Cell 33: 405-412.
13. DARNELL, J., LODISH, H., AND BALTIMORE, D. (1986). “Molecular Cell Biology,” Sci. Amer. Books, New York. DEAVEN, L. L., AND PETERSON, D. F. (1973). The chromosomes of CHO, an aneuploid Chinese hamster cell line: G-band, Cband, and autoradiographic analyses. Chromosoma 41: 129144.
NUCLEI
5.
REFERENCES 1.
CELL
14.
URLAUB, G., MITCHELL, P. J., KAS, E., CHASIN, L. A., FLINANAGE, V. L., MYODA, T. T., AND HAMLIN, J. (1986). Effect of gamma rays at the dihydrofolate reductase locus: Deletions and inversions. Somat. Cell Mol. Genet. 12: 5555566.