Demonstration of Apoptotic Cells in Tissue Sections by In Situ Hybridization Using Digoxigenin-labeled Poly(A) Oligonucleotide Probes to Detect Thymidine-rich DNA Sequences

Abstract

The recognition of apoptotic cells by morphological appearance alone may be difficult. We have investigated the use of in situ hybridization (ISH) with digoxigenin-labeled poly(A) probes to detect apoptotic cells in tissue sections. This method was compared to conventional morphologic assessment and in situ end-labeling (ISEL) in a range of tissues in which apoptosis is known to occur. ISH with poly(A) probes detected apoptotic nuclei in all tissues in which there was evidence of apoptosis as judged by conventional histology. ISH and, to a lesser extent, ISEL preferentially labeled shrunken but still intact nuclei with margination of chromatin, presumably at an early stage of apoptosis. The poly(A) hybridization was abolished by pretreatment of tissue sections with DNAse. After denaturation of DNA, poly(A) hybridized to nuclei in all cells. No convincing hybridization signal was detected in alcohol-fixed or fresh-frozen sections. Both ISEL and ISH labeled some of the nuclei in ischemic tissues. ISH with poly(A) oligonucleotide probes offers a simple alternative to ISEL for detection of cells in early stages of apoptosis. These probes hybridize to thymidine-rich sequences of DNA in the highly repeated Alu sequences within the nuclear genome. These sequences are believed to become available for hybridization after formalin fixation and paraffin embedding as a result of the apoptosis-related increase in the susceptibility of nuclear DNA to denaturation.

Keywords

apoptosis in situ hybridization poly(A)Alu sequences

Apoptosis is a form of programmed cell death, characterized morphologically by cell shrinkage, cytoplasmic blebbing, chromatin clumping, and formation of membrane-bound fragments of the nucleus, termed apoptotic bodies (Wyllie et al., 1984). Endonuclease activity cleaves DNA into oligosome-sized fragments that produce a DNA “ladder” on electrophoresis (Wyllie et al., 1984; Wyllie, 1980). Apoptosis is important for normal development and physiological cell turnover (Stewart, 1994; Wyllie, 1993) but also occurs in a number of pathological conditions, including human immunodeficiency virus infection (Gougeon and Montagnier, 1993) and malignancy, particularly after treatment by irradiation, chemotherapy, or hormone therapy (Hickman, 1992; Waters, 1992). Identification of apoptotic cells has assumed a key role in the investigation of normal development, tumorigenesis, tumor therapy, and a range of other conditions (Yoshiyama et al., 1994; Forloni et al., 1993).

The early morphological changes of apoptosis may be difficult to identify reliably under the light microscope. Apoptotic cells can be confused with infiltrating lymphocytes and cells in telophase (Wijsman et al., 1993). Apoptotic cells are often identified by in situ end-labeling (ISEL) of fragmented DNA (Gavrieli et al., 1992). Although ISEL has been widely used to study apoptosis, the method is not specific and may also detect necrotic cells and cells undergoing mitosis (Migheli et al., 1994; Gavrieli et al., 1992).

One feature of early apoptosis is an increase in the susceptibility of nuclear DNA to denaturation (Darzynkiewicz et al., 1992). The reasons for this are not fully understood. The internucleosomal cleavage of DNA is presumably a contributing factor, but in flow cytometry studies the susceptibility to denaturation seems to increase before demonstrable DNA cleavage (Hotz et al., 1994). This may allow parts of the genomic DNA to become available for hybridization, particularly the A-T-rich regions, which have lower melt temperatures than regions with a higher G-C content. A-T-rich regions are present in and adjacent to the long and short interspersed elements within the nuclear genome (Alu repeats). These are excellent targets for in situ hybridization, since many thousand Alu repeats occur in each copy of nuclear DNA (Lodish et al., 1995). We have investigated a novel method of detecting apoptotic cells by in situ hybridization (ISH) using poly(A) oligonucleotide probes to hybridize to poly(T) repeats associated with the Alu sequences.

Materials and Methods

We examined a wide range of tissues in which apoptosis is known to occur. These included tonsil, late secretory and menstrual endometrium, duodenum from a child with graft-vs-host disease, developing fetal hands, and brain tumors. Ischemic cerebral neocortex and hippocampus and infarcted cerebellum and myocardium were also examined. The material had been fixed in either 10% or 20% formalin and embedded in paraffin. ISH, ISEL, and conventional morphological assessment were used to identify apoptotic cells and the results were compared. RNAse and DNAse pretreatments were used to confirm the nature of the target sequences for ISH. The effects of the formalin fixation and paraffin embedding and of omission of salmon sperm DNA from the prehybridization solution were also investigated.

Tissue Fixation. Tissue was fixed in 10% neutral buffered formalin, 20% unbuffered formalin, or 70% ethanol for 1–5 days and then processed overnight, through alcohol and either xylene or chloroform, to paraffin.

Probes. Synthetic oligonucleotides 30 bases in length were synthesized on a DNA synthesizer. Similarly constructed poly(T) probes, and probes to the mitochondrial (Hilton et al., 1994) and Coxsackie virus (Hilton et al., 1992) genomes were used for control purposes. Terminal deoxynucleotidyl transferase was used to label the probes at the 3’ end with a homopolymer tail of dUTP-11-digoxigenin (Boehringer Mannheim; Mannheim, Germany). dATP, at a final concentration of 0.05 mM, was added to the labeling reaction of poly(A) to prevent self-annealing of labeled probe.

In Situ Hybridization. Paraffin sections 4 μm thick were collected on aminopropyltriethoxysilane-coated glass slides. The sections were dewaxed in xylene and graded alcohols, washed in 2 X SSC at 70°C for 10 min, treated with 2–10 μg/ml proteinase K for 60 min at 37°C, and refixed in 0.4% para-formaldehyde in 0.1 M phosphate buffer for 20 min at 4°C.

Frozen sections were also collected on aminopropyltriethoxysilane-coated glass slides. The sections were air-dried for 20–30 min within a cryostat at −20°C and then, in some cases, immersed in 20% unbuffered formalin for 30 min at room temperature.

The effects of DNAse and RNAse pretreatment were investigated by incubating sections in either DNAse I (Promega; Madison, WI) or RNAse A (Sigma; St Louis, MO) for 60 min at 37°C before hybridization. In other experiments, sections were denatured by heating to 95°C for 15 min on a hotplate.

Hybridization took place overnight at 37°C in 50 μl of buffer containing 0.1 ng/μl probe, 600 mM NaCl, 0.1 M phosphate buffer, 10% dextran sulfate and, except in certain experiments, 150 μg/ml sheared salmon sperm DNA. Post-hybridization washes consisted of 2 X SSC at 37°C. Thirty percent formamide was added to the hybridization solution and posthybridization washes with the probes to the mitochondrial and Coxsackie virus genomes. Sections were incubated for 10 min in Tris-HCl (pH 7.5)/0.5 M NaCl/3% bovine serum albumin, then for 30 min with anti-digoxigenin antibody (diluted 1:600) conjugated to alkaline phosphatase. Bound antibody was visualized by reaction with nitroblue tetrazolium salt and 5-bromo-4-chloro-3-indolyl phosphate solution with 1 μM levamisole, in the dark for up to 22 hr.

In Situ End-labeling. Paraffin sections 4 μm thick were collected on aminopropyltriethoxysilane-coated glass slides. The sections were dewaxed in xylene and graded alcohols, then rinsed in water and treated with 2–10 μg/ml proteinase K for 60 min at 37°C. Reaction buffer (50 μl) (Boehringer Mannheim) that included 5 U terminal deoxynucleotidyl transferase and 10 μM digoxigenin-11'-dUTP was added to the sections and left for 60 min at 37°C. Sections were incubated for 10 min in Tris-HCl (pH 7.5)/0.5M NaCl/3% bovine serum albumin, then for 30 min with anti-digoxigenin antibody (diluted 1:600) conjugated to alkaline phosphatase. Bound antibody was visualized by reaction with nitroblue tetrazolium salt and 5-bromo-4-chloro-3-indolyl phosphate solution containing 1 μM levamisole, in the dark for up to 2 hr.

Figure 1

High-magnification view of a tonsillar germinal center that contains apoptotic bodies (arrows), some within “tingible body” macrophages. Hematoxylin and eosin. Bar = 20 μm.

Assessment of Nuclear Size. The diameters of (a) normal nuclei, (b) nuclei and nuclear bodies identified as apoptotic in hematoxylin- and eosin-stained sections, (c) nuclei and nuclear bodies labeled by ISEL, and (d) nuclei and nuclear bodies labeled by ISH were compared in sections of a primitive neuroectodermal tumor composed of a relatively uniform population of tumor cells. The diameters were measured under a X 40 objective with the aid of an eyepiece graticule. Each data set comprised 100 measurements. Comparisons were made using Student's t-test.

Results

Conventional morphological assessment allowed relatively easy identification of the later stages of apoptosis, in which the nuclear chromatin becomes condensed and marginated and small apoptotic bodies form. Apoptotic cells were numerous in the basal epithelium and stroma of the late secretory and menstrual endometrium and within germinal centers of the tonsil (Figure 1). In the sections of tonsil, many of the apoptotic nuclei and bodies had been phagocytosed and could be identified in the cytoplasm of germinal center macrophages (“tingible body” macrophages). Scattered cells with apoptotic morphology were present in certain brain tumors (Figure 4a), in the duodenal mucosa in graft-vs-host disease, and in the subepidermal stroma of developing interdigital clefts in fetal hands. The sections of hippocampus, cerebellum, and myocardium showed the typical histological changes of acute ischemia and infarction.

ISH and ISEL produced nuclear signal in all of the locations in which cells showed histological changes of apoptosis (Figures 2-6), but both techniques detected only a proportion of the cells showing these changes and some apoptotic bodies remained unlabeled (Figures 4 and 6). Both techniques labeled intact, although rounded and shrunken, nuclei in those parts of the tissues that also contained fragmented nuclei and apoptotic bodies. The appearance of the labeled nuclei and the correspondence between the distribution of these and the smaller apoptotic bodies suggested that the former represented cells in relatively early stages of apoptosis. These cells were more consistently labeled by ISH than by ISEL. Conversely, ISEL appeared to detect more, although not all, of the smaller apoptotic bodies.

The diameter of nuclei labeled by ISEL or ISH was compared with that of normal and apoptotic nuclei as assessed in hematoxylin- and eosin-stained sections (Figure 7). The mean diameter of apoptotic nuclei was significantly smaller than that of normal nuclei (p<0.001). Both ISEL and ISH labeled nuclei of similar mean diameter to that of apoptotic nuclei in hematoxylin- and eosin-stained sections. Although there was extensive overlap between the size of nuclei labeled by the two molecular genetic methods, the mean diameter of those labeled by ISH was slightly larger than that of those labeled by ISEL, and this difference was statistically significant (p=0.002). Another difference between ISH and ISEL was evident in the sections of tonsil. ISEL yielded a diffuse signal in the cytoplasm of many germinal center macrophages, whereas ISH tended to produce a more discrete signal over the individual intracytoplasmic apoptotic nuclear bodies (Figures 2b and 3b).

Figure 2

(a) Low-power view of germinal center after ISEL. (b) Higher magnification reveals labeling of the nuclei of apoptotic cells (arrows) and diffuse cytoplasmic signal within many germinal center macrophages. Bars: a = 80 μm; b = 20 μm.

Figure 3

(a) Low-power view of germinal center after ISH with poly(A) oligonucleotide probe. (b) Relatively discrete labeling of apoptotic nuclei (arrows), some of which are within germinal center macrophages (arrowhead), is evident at higher magnification. Bars: a = 80 μm; b = 20 μm.

Figure 4

(a) Primitive neuroectodermal tumor, within which scattered cells are undergoing apoptosis (hematoxylin and eosin). (b) Nuclei labeled by the poly(A) probe tend to be rounded and smaller than those of the adjacent tumor cells, as would be expected of cells in the early stages of apoptosis. The smaller apoptotic bodies are not labeled. Bar = 20 μm.

Both ISH and ISEL labeled some of the nuclei in ischemic and infarcted tissues. This was particularly evident in sections of acutely infarcted cerebellum and in cerebral cortex showing morphological changes of acute ischemia.

Pretreatment of sections with DNAse I prevented the labeling of apoptotic nuclei by ISH. In contrast, RNAse A pretreatment had no effect on the nuclear signal. The possibility was considered that binding of probe to sheared, denatured salmon sperm DNA might be reducing the nuclear signal, but elimination of salmon sperm DNA from the prehybridization and hybridization buffers had no noticeable effect. Signal was slightly reduced but not abolished when the prehybridization 70°C wash in 2 X SSC was omitted.

To investigate the effects of the formalin fixation and the paraffin embedding, we performed ISH on unfixed and formalin-fixed cryostat sections of fresh tonsil and on paraffin sections of alcoholfixed tonsil. No convincing hybridization signal was detected in the fresh-frozen sections or in paraffin sections of alcohol-fixed tissue (Figure 8a). There was some labeling of apoptotic cells in the formalin-fixed frozen sections but this was weaker than in the paraffin-embedded material. Denaturation of DNA by heating paraffin sections of tonsil that had been fixed in alcohol or formalin to 95°C resulted in hybridization of poly(A) to all nuclei (Figure 8b).

Figure 5

Developing interdigital cleft in section through the hand of a 15-mm fetus. Poly(A) hybridization signal is visible over scattered nuclei in the subepidermal stroma. Bar = 80 μm.

Figure 6

High-magnification view of late secretory endometrium, within which poly(A)-labeled apoptotic cells are visible in an endometrial gland. The signal is strongest in the intact, albeit rounded and shrunken, apoptotic nuclei. Not all of the smaller apoptotic bodies are labeled. Bar = 15 μm.

Figure 7

Graphic representation of measurements of the diameter of normal nuclei, apoptotic nuclei, and nuclei labeled by ISEL or ISH in a primitive neuroectodermal tumor. Vertical lines indicate the range of measurements; the dot shows the mean, and the short horizontal bars ± 1 SEM. Both molecular genetic techniques label nuclei in a similar diameter range to those showing conventional morphological changes of apoptosis. However, the mean diameter of those labeled by ISH is slightly larger than of those labeled by ISEL.

Apoptotic cells were also identified by poly(T) hybridization, although in some sections this was partly obscured by strong cytoplasmic hybridization to the polyadenylated mRNA. Oligonucleotide probes to the mitochondrial and Coxsackie virus genomes did not produce any signal in apoptotic nuclei (Hilton et al., 1992, 1994).

Discussion

We have demonstrated that ISH with digoxigenin-labeled poly(A) oligonucleotide probes allows the detection of apoptotic cells in a wide range of tissues. The results are similar to those obtained with ISEL and, as with ISEL, not all cells showing morphological features of advanced apoptosis are labeled. Both methods preferentially label cells with intact, albeit shrunken, nuclei. In view of the shape and size of these labeled nuclei and the fact that they occur in the immediate vicinity of other cells containing apoptotic nuclear bodies, it appears likely that this labeling is of cells in a relatively early stage of apoptosis. ISH tends to label cells with slightly larger nuclei than does ISEL which, in turn, labels relatively more of the small apoptotic bodies. These findings are consistent with previous studies demonstrating that an increase in the susceptibility of nuclear DNA to denaturation occurs at an earlier stage of apoptosis than does DNA fragmentation (Hotz et al., 1994). This phenomenon may be partly attributable to the destabilizing effect of early proteolytic digestion of histones (Bruno et al., 1992). The reduction in the level of poly(A) hybridization signal during the later stages of apoptosis, after the formation of apoptotic bodies, may be due to progressive fragmentation of the genomic target sequences. The marked condensation of DNA within the small apoptotic bodies may also limit access of the poly(A) probe.

Figure 8

Alcohol-fixed, paraffin-embedded tonsil. (a) Absence of any poly(A) ISH signal. (b) Denaturation of DNA by heating the section to 95°C for 10 min has resulted in labeling of all of the nuclei with poly(A) oligonucleotides. Bar = 80 μm.

The poly(A) oligonucleotide probes hybridize to thymidine-rich nucleotide sequences. In the present study, the target sequences were shown to be within nuclear DNA rather than RNA. The intensity of signal in each labeled nucleus indicates the presence of large numbers of target sequences. A-T-rich regions occur in the highly repeated sequences of long and short (Alu family) interspersed elements of the human genome (Lodish et al., 1995) and are the most likely targets of the probes used in this study. These long and short interspersed elements are among the most abundant repeated sequences in the human genome, together accounting for approximately 10% of total human DNA. Although they may be transcribed they do not have any known function (Lodish et al., 1995). A-T-rich sequences have low melt temperatures and are therefore more susceptible to denaturation than other parts of the genome.

Previously, the increased susceptibility of apoptotic DNA to denaturation has been used to allow detection of apoptotic cells by flow cytometry (Hotz et al., 1994; Darzynkiewicz et al., 1992). Although the results of one study pointed towards an increase in the amount of single-stranded DNA in cells undergoing apoptosis (Darzynkiewicz et al., 1992), it is unlikely that single stranded DNA is present in significant amounts (Nuovo and Silverstein, 1988). Our results suggest that the hybridization of poly(A) and poly(T) to apoptotic cells in paraffin sections is facilitated by formalin fixation, which is known to lower the T_m of DNA (Nuovo and Silverstein, 1988), and by paraffin embedding, which involves heating of the tissue to 65°C for ∼3 hr. These factors presumably act in concert with the increase in the susceptibility of genomic DNA to denaturation, caused by the apoptosis itself.

ISH requires far less terminal transferase and digoxigenin-dUTP than does ISEL, so the cost per section is considerably lower. However, ISH is probably no more specific than ISEL, which has been reported to detect necrotic as well as apoptotic cells (Migheli et al., 1994; Gavrieli et al., 1992). The morphological features of apoptosis are distinctive, but many of the biochemical changes, including the increase in the susceptibility of nuclear DNA to denaturation and the formation of DNA “ladders,” may also occur in ischemic and toxic cell death (Ferrer et al., 1994; Fukuda et al., 1993; Darzynkiewicz et al., 1992). Whether these other forms of cell death should be regarded as apoptotic is debatable, but recent work suggests that some of the molecular determinants of ischemic cell death are similar to those involved in apoptosis (Graeber et al., 1996). Molecular techniques for assessment of apoptosis are therefore best used in conjunction with histological assessment.

Footnotes

Acknowledgements

This work was funded by a grant from the Edwin Luff Bequest. We also thank Howard Pringle, whose observation prompted this study.

References

Arends

Morris

Wyllie

(1990) Apoptosis. The role of the endonuclease. Am J Pathol 136:593–608

Bruno

Del Bino

Lassota

Giaretti

Darzynkiewicz

(1992) Inhibitors of proteases prevent endonucleolysis accompanying apoptotic cell death. Leukaemia 11:1113–1120

Darzynkiewicz

Bruno

Del Bino

Gorczyca

Hotz

Lassota

Traganos

(1992) Features of apoptotic cells measured by flow cytometry. Cytometry 13:795–808

Ferrer

Tortosa

Macaya

Sierra

Moreno

Munell

Blanco

Squier W

(1994) Evidence of nuclear DNA fragmentation following hypoxia-ischemia in the infant rat brain and transient forebrain ischemia in the adult gerbil. Brain Pathol 4:115–122

Forloni

Chiesa

Smiroldo

Verga

Salmona

Tagliavani

Angeretti

(1993) Apoptosis mediated neurotoxicity induced by chronic application of β amyloid fragment 25–35. Neuroreport 4:523–526

Fukuda

Kojiro

Chiu

(1993) Demonstration of extensive chromatin cleavage in transplanted Morris hepatoma 7777 tissue: apoptosis or necrosis? Am J Pathol 142:935–946

Gavrieli

Sherman

Ben-Sasson

(1992) Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol 119:493–501

Gougeon

Montagnier

(1993) Apoptosis in AIDS. Science 260: 1269–1270

Graeber

Osmanian

Jacks

Houseman

Koch

Lowe

Giaccia

(1996) Hypoxia-mediated selection of cells with diminished apoptotic potential in solid tumours. Nature 379:88–91

10.

Hickman

(1992) Apoptosis induced by anticancer drugs. Cancer Metastasis Rev 11:121–139

11.

Hilton

Day

Pringle

Fletcher

Chambers

(1992) Demonstration of the distribution of Coxsackie virus RNA in neonatal mice by non-isotopic in situ hybridisation. J Virol Methods 40:155–162

12.

Hilton

Love

Goodwin

Pringle

(1994) Demonstration of mitochondrial ribosomal RNA in frozen and paraffin-embedded sections of skeletal muscle by in situ hybridisation. Neuropathol Appl Neurobiol 20:573–576

13.

Hotz

Gong

Traganos

Darynkiewicz

(1994) Flow cytometric detection of apoptosis: comparison of assays in situ DNA degradation and chromatin changes. Cytometry 15:237–244

14.

Lodish

Baltimore

Berk

Zipursky

Matsudaira

Darnell

(1995) The molecular anatomy of genes and chromosomes. In Molecular Cell Biology. New York, WH Freeman, 334–338

15.

Migheli

Cavalla

Marino

Schiffer

(1994) A study of apoptosis in normal and pathological nervous tissue after in situ end-labeling of DNA strand breaks. J Neuropathol Exp Neurol 53:606–616

16.

Nuovo

Sliverstein

(1988) Comparison of formalin, buffered formalin and Bouin's fixation on the detection of human papillomavirus deoxyribonucleic acid from genital lesions. Lab Invest 59:720–724

17.

Stewart

(1994) Mechanisms of apoptosis: Integration of genetic, biochemical and cellular interactions. J Natl Cancer Inst 86:1286–1296

18.

Waters

(1992) Radiation-induced apoptosis in a murine T-cell hybridoma. Cancer Res 52:883–890

19.

Wijsman

Jonker

Keijzer

Van de Velde

CJH

Cornelisse

Van Dierendonck

(1993) A new method to detect apoptosis in paraffin sections: in situ end-labeling of fragmented DNA. J Histochem Cytochem 41:7–12

20.

Wyllie

(1993) Apoptosis. Br J Cancer 67:205–208

21.

Wyllie

(1980) Glucocorticoid-induced thymocyte apoptosis is associated with endogenous endonuclease activation. Nature 284:555–556

22.

Wyllie

Morris

Smith

Dunlop

(1984) Chromatin cleavage in apoptosis: association with condensed chromatin morphology and dependence on macromolecular synthesis. J Pathol 142:67–77

23.

Yoshiyama

Yamada

Asanuma

Asahi

(1994) Apoptosis related antigen, Le^v and nick-end labeling are positive in spinal motor neurons in amyotrophic lateral sclerosis. Acta Neuropathol (Berl) 88:207–211