Mutation Research 473 (2001) 163–168

Evidence of abundant constitutive alkali-labile sites in human5 bp classical satellite DNA loci by DBD-FISH

José Luis Fernándeza,∗, Fernando Vázquez-Gundına,Marıa Teresa Riveroa, Vicente Goyanesb, Jaime Gosálvezc

a Laboratorio de Genética Molecular y Radiobiologıa, Centro Oncológico de Galicia, 15009 La Coruña, Spainb Sección de Genética, Hospital “Teresa Herrera”, Instituto de Ciencias de la Salud, Universidad de La Coruña, La Coruña, Spain

c Unidad de Genética, Facultad de Biologıa, Universidad Autónoma de Madrid, Madrid, Spain

Received 14 June 2000; received in revised form 18 October 2000; accepted 19 October 2000

Abstract

Human blood leukocytes within an agarose matrix were deproteinized and exposed to an alkaline denaturation that generatessingle-stranded DNA (ssDNA) starting from the ends of spontaneous basal DNA breaks and alkali-labile sites. Since the amountof ssDNA produced within a specific sequence area may be detected by hybridization with a specific probe, we quantified thisin situ in different satellite DNA loci (DBD-FISH: DNA Breakage Detection FISH). The DBD-FISH signal, corrected for therespective FISH signals in metaphase, was remarkably strong in the 5 bp classical satellite DNA domains analyzed (D1Z1,D9Z3, DYZ1), intermediate in the classical satellite 1 DNA sequences, and low in the alphoid satellite regions (D1Z5, DXZ1,all centromeres). This result is evidence of a high density of constitutive alkali-labile sites, probably abasic sites, within the5 bp satellite DNA sequences in human blood leukocytes. The presence and relative abundance of alkali-labile sites couldexplain the high frequency of spontaneous breakage and rearrangements in pericentromeric heterochromatin of chromosomes1, 9, and 16, but not in Yqh, when this chromatin is undercondensed through spontaneous or induced demethylation, i.e. ICFsyndrome or 5-azacytidine treatment. © 2001 Elsevier Science B.V. All rights reserved.

Keywords:DBD-FISH; Alkali-labile sites; Human satellite DNA; Classical satellite DNA; ICF syndrome; 5-Azacytidine

1. Introduction

Human satellite DNA includes a number of dif-ferent tandemly repeated DNA sequence families,organized in very large arrays. Alpha satellite DNA,located in the centromeric region of all chromosomes,is composed of monomers of 171 bp, organized inhigher-order repeated units. Their sequence diver-gence allows specific subfamilies for each of the hu-man chromosomes to be isolated [1]. Classical satellite1, which is AT-rich, is composed of two alternating

∗ Corresponding author. Fax:+34-981-287-122.E-mail address:genetica@cog.es (J.L. Fernandez).

units of 17 and 25 bp, respectively, that are tandemlyrepeated. It appears to be located on pericentromericregions of chromosomes 3 and 4, and on pericen-tromeric and distal short arm regions of the acrocentricchromosomes [2]. The 5 bp classical satellites 2 and 3contain sequence arrays based on tandem repetitionsof variants of the “core” sequence 5′-TTCCA-3′.Satellite 2 is believed to be a degenerate and morerecently diverged form of satellite 3. Specific sub-families are present in chromosome-specific domains,being more abundant in the juxtacentromeric hete-rochromatin of chromosomes 1, 9, 16 and in the longarm of the Y chromosome [3,4]. The aforementionedspecific families, besides others (e.g. Tyler-Smith &

0027-5107/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S0027-5107(00)00146-9

164 J.L. Fernandez et al. / Mutation Research 473 (2001) 163–168

Willard [1]) differ not only at the DNA level in termsof base sequence, repeat unit, higher-order arrays ofrepeated units, overall length and specific chromo-some positions of the tandem blocks, but also possiblyin protein composition and three-dimensional chro-matin folding superstructure [5]. The latter has beenconcluded through quantitative FISH studies after insitu digestion with nucleases [6,7].

We have recently developed a technique for insitu detection of DNA strand breaks within specificsequence regions in single cells (DBD-FISH: DNABreakage Detection-FISH, Fernández et al. [8]). Cellstrapped within an agarose microgel are introducedinto an alkaline unwinding solution, deproteinized inlysing solutions and dehydrated in increasing ethanolbaths. The alkali transforms strand breaks into ss-DNA, which serves as a target for hybridization ofDNA probes. The specific probe selects the chromatinarea to be analyzed. As DNA breaks increase withina specific target region (e.g. by ionizing radiation ex-posure), more ssDNA is produced by the alkali in thelocus and more probe hybridizes, subsequently giv-ing rise to a more intense FISH signal, which is thenquantified using a digital image analysis system. Notonly DNA strand breaks but also alkali-labile sitesmay be detected using DBD-FISH. In this report, wepresent the results of the application of DBD-FISH inthe analysis of the background labelling within spe-cific satellite DNA sequence areas from human bloodcells. Surprisingly, a remarkably strong labelling wasevident in the 5 bp classical satellite (D1Z1 locus,satellite 2; D9Z3 and DYZ1 loci, satellite 3) com-pared with classical satellite 1 and alphoid satellites(D1Z5, DXZ1 and all centromeres loci).

2. Materials and methods

DBD-FISH was performed as described by Fernán-dez et al. [8] with variations. The entire study wascarried out on leukocytes from the same male donor.Buffy coat was mixed with low melting point agarosewith a final concentration of 0.7% at 37◦C. 50ml of themixture were deposited onto a glass slide precoatedwith 0.65% standard agarose dried at 80◦C, coveredwith a coverslip (24 mm× 60 mm) and allowed to so-lidify at 4◦C. Coverslips were carefully removed andslides were introduced sequentially into lysing solu-

tions 1 (0.4 M Tris-HCl, 0.8 M DTT, 1% SDS) and 2(0.4 M Tris-HCl, 2 M NaCl, 1% SDS) both at pH 7.5at room temperature, for 20 and 10 min, respectively.After protein removal, nucleoids were washed in 0.9%NaCl for 15 min and then incubated in the alkaline un-winding solution (0.03 M NaOH, 1 M NaCl, pH 12.2)for only 150 s at 7◦C. This produces ssDNA motifsstarting from the ends of spontaneous strand breaks oralkali-labile sites. Conditions of unwinding were cho-sen so that, whereas the diffusion of alkali affects allcells homogeneously within the gel, the extension ofmelted ssDNA produced starting from a break is re-stricted as far as possible, so that the DNA breakagequantified is mainly that in the specific locus probed.After neutralizing with 0.4 M Tris, pH 7.5, 15 min,to stop denaturation, nucleoids were washed in 0.9%NaCl for 5 min, dehydrated by sequential 70, 90 and100% ethanol baths, 2 min each, and air-dried.

In another experimental variant, embedded cellswere first subjected to the alkaline unwinding solutionand then neutralized and deproteinized in the lysingsolutions. After washing, they were dehydrated andair-dried.

Biotin-labelled probes were denatured and incu-bated overnight on dried slides. Alphoid probes ofchromosome 1 (D1Z5), chromosome X (DXZ1), acocktail of alphoid sequences for all centromeres, aswell as specific probes for 5 bp classical satellite 2of chromosome 1 (D1Z1) and satellite 3 of chro-mosomes 9 (D9Z3, formerly D9Z1) and Y (DYZ1),were also hybridized. All these probes were purchasedfrom Appligene Oncor Cytocell. A classical satellite 1oligonucleotide probe (S.1) was synthesized and usedas described by Tagarro et al. [9]. After washings, bi-otinylated probes were detected with streptavidin-Cy3(Sigma), without signal amplification. Counterstainingwas performed using DAPI (1mg/ml) in Vectashield(Vector).

Signals were acquired in similar conditions ofcapture time and resolution, with a high sensitivityCCD camera (Ultrapix 1600, Astrocam) coupled to aDMRB epifluorescence microscope (Leica). Groupsof 55 images per experimental point were stored andcorrected for electronic noise and non-uniform sampleillumination. The images were exported from apf. toimg. files, maintaining the grey levels, and analyzedusing a macro designed with Visilog 5.1 software(Noesis). This allows for thresholding, background

J.L. Fernandez et al. / Mutation Research 473 (2001) 163–168 165

subtraction, and measures the whole fluorescenceintensity (WFI: surface area× mean fluorescenceintensity) of the signals.

DBD-FISH signals from a specific locus were rela-tivized to the FISH signal of the same locus obtainedin 15–20 conventional denatured metaphases from thesame individual, all with a similar condensation level.This allows the comparison of the DBD-FISH signalsamong the different satellite DNA loci, thus correctingfor the differences in the amount of target sequence,possible variations in the biotinylation level of eachprobe, and other possible sources of distortion.

3. Results

When human blood leukocytes embedded in anagarose matrix are exposed to lysing solutions, thechromatin is depleted of proteins producing nucleoids,which are then exposed to the alkali denaturationstep. The alkaline unwinding solution generates ss-DNA that may be detected by FISH. The densityof ssDNA produced in different satellite DNA se-quences was quantified, demonstrating differencesin alkali-sensitivity among them (Fig. 1, Table 1DBD-FISH L → U). When analyzing alphoid D1Z5and DXZ1 loci, only 30 and 45% of cells, respec-tively, produced very small and faint backgroundsignals. All cells appeared labelled when probing allcentromeres, satellite 1 and 5 bp satellites. No signalwas evident at any locus if the alkaline denaturationstep was suppressed. It was clearly evident that 5 bpsatellite sequence areas are remarkably more sensitiveto alkali denaturation than alphoid sequence areas,satellite 1 having an intermediate sensitivity. Statis-tically significant differences were found among allloci (one-way ANOVA,P < 0.05), except betweenD1Z5 and DXZ1, probably due to the very low pro-portion of cells with signal. All centromeres showed ahigher relative signal than D1Z5 and DXZ1, probablyreflecting differences in sensitivity among differentalphoid loci, and always within a low range. The5 bp satellite sequences also showed different sensi-tivities, within a high range, the DYZ1 locus beingthe least affected by alkali denaturation. The highalkali-sensitivity of these different 5 bp satellite DNAsequences was confirmed in leukocytes from anotherthree women and two men.

If leukocytes were first exposed to alkali denatu-ration and then to deproteinization, the differences inthe density of ssDNA generated among the differentloci were maintained, though at a lower scale (Table 1,DBD-FISH U → L). This was due to the interfer-ence exerted by cytoplasm and DNA-bound proteinsto the accessibility of alkali to DNA. This approxi-mately seven-fold protective effect seemed more orless homogeneous among the different loci (Table 1,L → U/U → L). All experiments were replicatedthree times and gave similar results. Furthermore, nodifferences were found between isolated lymphocytesand granulocytes (data not shown).

4. Discussion

The differences in the amount of ssDNA generatedby the alkali within the different satellite DNA loci arenot dependent on the different higher-order chromatinorganizations, since these are destroyed by the lysingstep preceding the alkali incubation. These differencesare likely to be related to the basic DNA structure.DBD-FISH detects DNA strand breaks, since the al-kali unwinding techniques of current use in radiobi-ology to measure DNA breaks are based on the factthat the alkali generates ssDNA motifs beginning fromthe ends of a DNA break [10]. In our case, thoughDNA breaks are spontaneously produced in normalliving cells, the extreme background labelling in 5 bpsatellites is difficult to explain by extensive nativeDNA fragmentation. The fact that this remarkable la-belling is also produced when living cells are imme-diately exposed to alkali before the lysing step, ar-gues against the extensive induction of DNA breaksduring the incubation in lysing solutions. Alkali notonly transforms DNA breaks into ssDNA, but alsoalkali-labile sites, i.e. DNA lesions which are con-verted into DNA strand breaks upon alkali treatment.They may be either spontaneous or induced, e.g. byionizing radiation exposure, corresponding chemicallyto abasic sites or altered deoxyriboses [11]. Classical5 bp satellite DNAs should contain a large amount ofconstitutive alkali-labile sites, which explains the re-markable level of DBD-FISH labelling. These may beabasic sites better than damaged sugars, the latter be-ing more specific to radical damage [11]. Constitutiveabundant alkali-labile sites have been described in the

166 J.L. Fernandez et al. / Mutation Research 473 (2001) 163–168

Fig. 1. Background DBD-FISH signals in human blood leukocytes accompanied by the respective FISH signals in conventionally denaturedlymphocyte metaphases from the same individual, in different satellite DNA sequences; a, a′: alphoid D1Z5 locus; b, b′: alphoid DXZ1locus; c, c′: all centromeric alphoid loci; d, d′: classical satellite 1 loci; e, e′: classical 5 bp satellite DYZ1 locus; f, f′: classical 5 bpsatellite D9Z3 locus; g, g′: classical 5 bp satellite D1Z1 locus. Taking into account the amount of sequence in metaphase, the classical 5 bpsatellite DNA loci show a remarkably stronger basal DBD-FISH labelling than alphoid sequences, classical satellite 1 being intermediate.

J.L. Fernandez et al. / Mutation Research 473 (2001) 163–168 167

Table 1Whole fluorescence intensity (WFI) values (mean± S.D.) of the FISH signals from different satellite DNA sequences in conventionalmetaphase spreads and after DBD-FISH on first lysed and then unwound cells (L→ U) or first unwound and then lysed cells (U→ L)a

Satellite loci Fish in metaphase DBD-FISH (L→ U) DBD-FISH (U → L) (L → U)/(U → L)

Alphoid D1Z5 7947.63±3891.28 81.46±144.70 12.24±51.91 6.65DXZ1 18818.80±3788.97 121.19±162.48 17.91±55.55 6.77All centromeres 1480327.00±287382.51 325.69±106.27 55.30±18.52 5.89

Satellite 1 276876.95±52728.56 383.70±158.91 75.85±38.53 5.06

5 bp satellite DYZ1 48023.16±5316.08 816.48±183.60 93.49±42.71 8.73D9Z3 442828.83±98564.05 1943.40±536.29 254.37±115.72 7.64D1Z1 734927.80±238647.73 2353.68±1007.66 347.75±185.76 6.77

a DBD-FISH values were corrected for the respective FISH signals in metaphase (WFI DBD-FISH/WFI FISH metaphase× 1000).

whole genome of human and mouse sperm, chickenerythrocytes and in mouse kidney DNA. Evidence forthem came from the alkali treatment of these cellsfor single-cell microgel electrophoresis (comet tail),alkaline agarose gel electrophoresis, or alkaline elu-tion assays [12–14]. These authors argued that thesesites are related to the characteristic tight packing ofchromatin in these cells. If this is true, the presenceof abundant constitutive alkali-labile sites within 5 bpsatellite DNA sequences may also be involved in theproduction of a special DNA compaction characteris-tic of these heterochromatic loci.

Human satellite DNA families differ in their DNAsequence, protein composition, higher-order chro-matin folding, and functionalism [5]. Thus, whilealphoid loci may be involved in mitotic segregation, asdemonstrated when generating human artificial chro-mosomes [15], 5 bp classical satellite DNA loci donot have such a clear role. Furthermore, alphoid lociare not decondensed by treatment with 5-azacytidine(5-azaC), unlike 5 bp satellite loci [16]. The demethy-lation effect of 5-azaC specifically within 5 bp classi-cal satellite DNA domains could lead to the loss ofmethyl-specific protein complexes that stabilize a par-ticular higher-order chromatin structure, resulting inuncoiling. It favours somatic associations, and break-age and rearrangements specifically among these loci[17]. The same phenomena of decondensation andinstability involving 5 bp satellite loci were describedin some lymphoblastoid cell lines with spontaneoushypomethylation of these specific sequences [18], andin the ICF syndrome. In this later case, mutations ofDNM3B de novo methyltransferase result in loss ofmethylation of 5 bp satellite DNA sequences, leading

to decondensation and recombination in lymphocytechromosomes [19–22]. Though demethylation, eitherspontaneous or induced, promotes decondensationand somatic association of large chromatin blockscontaining 5 bp satellites, this alone may not explaintheir spontaneous breakage and rejoining. Possibly,the presence of extensive alkali-labile sites specifi-cally in these sequences could promote their breakageand interchange through the repair machinery, whensomatic paired after decondensation. In fact, thoughall 5 bp satellite heterochromatic blocks of chromo-somes 1, 9, 16 and Y appear hypomethylated andundercondensed both in the ICF syndrome and after5-azaC treatment, breakage and rearrangements wereonly described involving the 1, 9, 16, but not the Ychromosome [17,23–26]. No explanation has beenfound for this difference. Nevertheless, there is a cor-relation between this and the density of alkali-labilesites detected by DBD-FISH, the DYZ1 locus be-ing in the very low range within the 5 bp satellites(Table 1). This might also explain the preferentialinvolvement of satellite 3 DNA in chromosomal rear-rangements that lead to Robertsonian translocations[27] in human acrocentric chromosomes.

Acknowledgements

We are grateful to Asunción Campos, Susana Do-rado and Belén Ferreiro for their excellent technicalassistance. We also thank Dr. Phil Mason for improv-ing the English style of the manuscript. This workwas supported by the Consejo de Seguridad Nuclear(CSN), Spain (grant SPR/317/99/640.000).

168 J.L. Fernandez et al. / Mutation Research 473 (2001) 163–168

References

[1] C. Tyler-Smith, H.F. Willard, Mammalian chromosomestructure, Curr. Opin. Genet. Dev. 3 (1993) 390–397.

[2] J. Meyne, E.H. Goodwin, R.K. Moyzis, Chromosomelocalization and orientation of the simple sequence repeat ofhuman satellite I DNA, Chromosoma 103 (1994) 99–103.

[3] R.K. Moyzis, K.L. Albright, M.F. Bartholdi, L.S.Cram, L.L. Deaven, C.E. Hildebrand, E.N. Joste, J.L.Longmire, J. Meyne, T. Schwarzacher-Robinson, Humanchromosome-specific repetitive DNA sequences: novelmarkers for genetic analysis, Chromosoma 95 (1987) 375–386.

[4] M. Jeanpierre, Human satellite 2 and 3, Ann. Génét. 37 (1994)163–171.

[5] P. Vogt, Code domains in tandem repetitive DNA sequencestructures, Chromosoma 101 (1992) 585–589.

[6] J.L. Fernández, D. Valverde, V. Goyanes, I. Buño, J. Gosálvez,Alu I in situ digestion of human alphoid and classical satelliteDNA regions: high-resolution digital image analysis of FISHsignals from condensed and extended chromatin, Cytogenet.Cell Genet. 76 (1997) 94–100.

[7] J. Gosálvez, C. López-Fernández, V. Goyanes, R. Mezzanotte,Chromosome differentiation using nucleases: an overview,in: N. Henriques-Gil, J.S. Parker, M.J. Puertas (Eds.),Chromosomes Today, Vol. 12, Chapman and Hall, London,1997, pp. 23–49.

[8] J.L. Fernández, V. Goyanes, J. Ramiro-Diaz, J. Gosálvez,Application of FISH for in situ detection and quantificationof DNA breaks, Cytogenet. Cell Genet. 82 (1998) 251–256.

[9] I. Tagarro, J. Wiegant, A.K. Raap, J.J. González-Aguilera,A.M. Fenández-Peralta, Assignment of human satellite 1DNA as revealed by fluorescence in situ hybridization witholigonucleotides, Hum. Genet. 93 (1994) 125–128.

[10] B. Rydberg, The rate of strand separation in alkali of DNA ofirradiated mammalian cells, Radiat. Res. 61 (1975) 274–287.

[11] C. Von Sonntag, The Chemical Basis of Radiation Biology,Taylor and Francis, London, 1987.

[12] D.B. Singh, R.R. Danner, M.T. Tice, G.D. McCoy, E.L.Collins, Schneider, Abundant alkali-sensitive sites in DNA ofhuman and mouse sperm, Exp. Cell Res. 184 (1989) 461–470.

[13] D.W. Fairbain, W.A. Reyes, K.L. O’Neill, Alkali-labile sitesare prevalent in kidney tissue DNA, Cancer Lett. 81 (1994)67–76.

[14] G. Haines, B. Marples, P. Daniel, I. Morris, DNA damagein human and mouse spermatozoa after in vitro-irradiationassessed by the comet assay, Adv. Exp. Med. Biol. 444 (1998)79–91.

[15] J.J. Harrington, G. Van Bokkelen, R.W. Mays, K. Gustashaw,H.F. Willard, Formation of de novo centromeres andconstruction of first-generation human artificial micro-chromosomes, Nature Genet. 15 (1997) 345–355.

[16] J.L. Fernández, V. Goyanes, S. Pereira, C. López-Fernández,J. Gosálvez, 5-azacytidine produces differential underconden-sation of alpha, beta and classical human satellite DNAs,Chrom. Res. 2 (1994) 29–35.

[17] N. Kokalj-Vokac, A. Almeida, E. Viegas-Péquignot, M.Jeanpierre, B. Malfoy, B. Dutrillaux, Specific induction ofuncoiling and recombination by azacytidine in classicalsatellite-containing constitutive heterochromatin, Cytogenet.Cell Genet. 63 (1993) 11–15.

[18] A. Almeida, N. Kokalj-Vokac, D. Lefrançois, E.Viegas-Péquignot, M. Jeanpierre, B. Dutrillaux, B. Malfoy,Hypomethylation of classical satellite DNA and chromosomeinstability in lymphoblastoid cell lines, Hum. Genet. 91 (1993)538–546.

[19] M. Jeanpierre, C. Turleau, A. Aurias, M. Prieur, F. Ledeist, A.Fischer, E. Viegas-Péquignot, An embrionic-like methylationpattern of classical satellite DNA is observed in ICFsyndrome, Hum. Mol. Genet. 2 (1993) 731–735.

[20] G.L. Xu, T.H. Bestor, D. Bourc’his, C.L. Hsieh,N. Tommerup, M. Bugge, M. Hultén, X. Qu, J.Russo, E. Viegas-Péquignot, Chromosome instability andimmunodeficiency syndrome caused by mutations in a DNAmethyltransferase gene, Nature 402 (1999) 187–191.

[21] R.S. Hansen, C. Wijmenga, P. Luo, A.N. Stanek, T.K.Canfield, C.N. Weemaes, S.M. Gartler, The DNMT3BDNA methyltransferase gene is mutated in the ICFimmunodeficiency syndrome, Proc. Natl. Acad. Sci. U.S.A.96 (1999) 14412–14417.

[22] M. Okano, D.W. Bell, D.A. Haber, E. Li, DNAmethyltransferases Dnmt3a and Dnmt3b are essential forde novo methylation and mammalian development, Cell 99(1999) 247–257.

[23] D.F.C.M. Smeets, U. Moog, C.M.R. Weemaes, G.Vaes-Peeters, G.F. Merkx, J.P. Niehof, G. Hamers, ICFsyndrome: a new case and review of the literature, Hum.Genet. 94 (1994) 240–246.

[24] D.C. Brown, E. Grace, A.T. Sumner, A.T. Edmunds,P.M. Ellis, ICF syndrome (immunodeficiency, centro-meric instability and facial anomalies) investigation ofheterochromatin abnormalities and review of clinical outcome,Hum. Genet. 96 (1995) 411–416.

[25] M. Stacey, M.S. Bennett, M. Hultén, FISH analysis onspontaneously arising micronuclei in the ICF syndrome, J.Med. Genet. 32 (1995) 502–508.

[26] A.T. Sumner, A.R. Mitchell, P.M. Ellis, A FISH studyof chromosome fusion in the ICF syndrome: involvementof paracentric heterochromatin but not of the centromeresthemselves, J. Med. Genet. 35 (1998) 833–835.

[27] C.H. Gravholt, U. Friedrich, M. Caprani, A.L. Jørgensen,Breakpoints in Robertsonian translocations are localized tosatellite III DNA by fluorescence in situ hybridization,Genomics 14 (1992) 924–930.