Analysis of Human Mitochondrial Transcripts Using Electron Microscopic in Situ Hybridization

Analysis of human mitochondrial transcripts using electron microscopic in situ hybridization

FRANCOISE ESCAIG-HAYE1'2, VLADIMIR GRIGORIEV3, GABRIEL PERANZI4, PATRICK LESTIENNE5 and JEAN-GUY FOURNIER12

1INSERM Uniti 43, Hdpital St Vincent de Paul, 74 avenue Denfert-Rochereau, 75674 Paris Cedex 14, France 2INSERM Unit6 153, 17 rue du Fer-a-Moulin, 75005 Paris, France 3Iuanousky Institute, Gamaleya Street, Moscow, USSR 4INSERM UnM 10, Hdpital Bichat, 170 boulevard Ney, 75877 Paris Cedex 18, France 5INSERM Unite 298, CHR, 49033 Angers Cedex, France

Summary

Human mitochondrial transcripts have been exam- analysis indicates that the mitochondria concentrate ined at the ultrastructural level. After contact with the labeling with intensities that vary with the type of ultrathin sections of a human lymphoid cell line RNA and that the nucleus induces a light hybridiz- (CEM) embedded in Lowicryl K4M, biotinylated ation signal with each mitochondrial probe. The mitochondrial probes yield specific hybrids ident- visualization of human mitochondrial DNA ex- ified by a colloidal gold immunocytochemistry pression in correlation with the fine anatomy of the marker that visualizes rRNA and mRNA coding for mitochondria constitutes a new approach for funda- respiratory chain polypeptides CO II, CO III and mental research on the organelle and for analyzing ATPase-6. The mitochondrial transcripts are prefer- its behaviour in human mitochondrial diseases. entially located close to the inner membrane, particularly the cristae, suggesting that intra-organelle protein synthesis is intimately associated with the Key words: mitochondria, transcription, ultrastructure, mitochondrial membrane system. Quantitative hybridization.

Introduction and function, since the morphological organization of mitochondria is visible uniquely under the electron The mitochondria occupy a unique situation among the microscope. Recently, several attempts have been made to cellular organelles because they play a major role in the adapt an in situ hybridization protocol to electron cell respiratory function and this function requires microscopy. Among the different approaches, those using coordinated interaction of the nuclear and the mitochon- probes non-isotopically labeled with biotin (Langer et al. drial genomes. The recent application of recombinant 1981; Singer and Ward, 1982; Singer et al. 1987) to perform DNA technology to the human mitochondrial genome has hybridization either before (Wolber et al. 1989; Silva et al. permitted the determination of its complete sequences and 1989; Singer et al. 19896), after (Binder et al. 1986; to observe the extreme compactness of its organization Webster et al. 1987; Brangeon et al. 1989; Escaig-Haye et (Anderson et al. 1981). The heavy strand contains the al. 1989; Thiry and Thiry-Blaise, 1989) or without (Radic et genes for two rRNAs, 14 tRNAs and 12 reading frames, al. 1987; Singer et al. 1989a) cell embedding in resin, offer while the light strand contains the genes for eight tRNAs a very rapid and effective method for the detection of the and one reading frame. The two strands are transcribed in nucleic acid integrated in their sub-cellular environment. the form of polycistronic molecules that are processed by In this work, the method of approach is to hybridize endonucleotide cleavage to yield the individual mature ultra-thin sections of material embedded in the hydrosol- mRNA species necessary for the synthesis of 13 polypep- uble resin Lowicryl K4M with a biotinylated DNA probe, tide components of the respiratory chain (Murphy et al. then, using colloidal gold immunocytochemistry marker 1975; Qjala et al. 1981; Bogenhagen et al. 1984; Chang and technology, to reveal the site of the specific hybrid Clayton, 1984; Shuey and Attardi, 1985; Kruse et al. 1989). formation. We have previously employed this protocol for To gain insight into the genetic functions of human detecting cellular RNA (Escaig-Haye et al. 1989) and a mitochondria required for oxidative phosphorylation, it similar approach has been used to detect ribosomal appeared essential to examine the mitochondrial tran- mitochondrial RNA in Drosophila cells (Binder et al. scripts in relation to the fine structure of the organelle's 1986). In the present case, expression of the human morphology. In that case, ultrastructural detection of mitochondrial genome was examined in a lymphoblastoid nucleic acid molecules is the only means of examining the human cell line with three different mitochondrial probes, molecular aspect of the relationship between structure one detecting ribosomal RNA, another detecting simul-

Journal of Cell Science 100, 851-862 (1991) Printed in Great Britain © The Company of Biologists Limited 1991 851 taneously two mRNA populations corresponding to cyto- floating the grids on drops of the hybridization solution 1 chrome c oxidase subunit III (CO III) mRNA and ATPase containing the DNA probes at a concentration of 10-20 ng ml" in subunit 6 (ATPase-6) mRNA, and the last one detecting a moist chamber for 16 h at 37 °C. The grids were then washed cytochrome c oxidase subunit II (CO II) mRNA. successively in 50% formamide, 4xSSC (twice for 5min), 4xSSC (twice for 5 min), 0.2xSSC (twice for 5 min) and distilled water Qualitative analysis of the results indicated that the (once for 2 min). mitochondrial RNA was mainly localized close to the inner membrane of the mitochondria, particularly the cristae; Immunocytochemistry this was confirmed by quantitative analysis, which also showed that very different amounts of molecules were After hybridization washings, grids were incubated in PBS buffer containing 1 % bovine serum albumin (PBS/BSA, 1 %) for 15 min. detected according to the probe used. Several controls were Then the sections were incubated with rabbit antibiotin (Enzo, performed to confirm the specificity of the data and Biochemical) diluted at 1 % in PBS/BSA (1 %) for 60 min. After statistical analysis indicated that the nucleus reacted with incubation with the antibody, grids were rinsed in PBS buffer the mitochondrial probes. containing 0.05% Tween 20 (PBS/Tween 20, 0.05%) twice for 10 min and then were kept in contact for 60 min with goat anti- rabbit antibody complexed with 10 nm gold particles (Janssen, Materials and methods Belgium) diluted at 1/50 in PBS/BSA (1%) solution. Sub- sequently, grids were rinsed in PBS/Tween 20, 0.05% and Materials distilled water twice for 10 min each. Sections were next stained with uranyl acetate for 20 min and rinsed in distilled water twice The CEM lymphoblastoid cell line was maintained in suspension for 5 min. They were observed in a CM10 Philips electron in RPMI medium supplemented with 10% fetal calf serum, microscope at an acceleration voltage of 60 kV. streptamicin and penicillin. Human peripheral blood mononuclear cells were purified on a Ficoll-Hypaque gradient and stimulated for 3 days with the same medium containing PHA Quantitation analysis of labeling density (lO/igml"1). Human tissues were obtained from lymph-node and The labeling densities over various cell compartments of the CEM muscle biopsies and from a surgically removed appendix. cells were evaluated on negative electron micrographs taken at a minimal magnification of x 11500. A total of 15-25 electron micrographs were recorded for each sample. The surface areas Cells and tissue processing 3 occupied by the mitochondria and nucleus were evaluated using Cell pellets and 1 mm pieces of tissue were fixed by immersion for an image-analysing system, Mode Biocom 200. The surface area 2 h either in 4 % paraformaldehyde and 0.1% glutaraldehyde or in of the cytoplasm was obtained by subtracting the mitochondria 2 % paraformaldehyde and 0.2 % glutaraldehyde, diluted in 0.1 M and nucleus areas from the total surface of the negative electron sodium phosphate buffer (pH7.4), and then rinsed in the buffer micrographs. The number of gold particles present over each of twice for 5 min each. The biological material could then be stored the compartments was counted manually and the density of in the buffer complemented with sucrose for several days at 4°C. labeling was calculated by dividing the number of particles by the Embedding in Lowicryl was done according to Bendayan (1984), surface area. The results are expressed as the mean number of as previously described (Grigoriev et al. 1989; Escaig-Haye et al. gold particles^m~2±standard deviation of different samples. 1989). Ultrathin sections (60-80 nm) were cut on a Reichert Ultracut ultramicrotome and sections were collected directly on Statistical analysis was performed by Student's i-test. acetone-treated naked gold grids (hexagonal 600 mesh). Grids were kept at room temperature for no more than 3 weeks before use. Results

Probes and nick-translation To analyse the human mitochondrial transcripts at the The mitochondrial probes consist of three different fragments of ultrastructural level, probes specific for different mito- the DNA genome. The first corresponds to 1.5 kb of Xbal 1-2 chondrial genes were hybridized to ultrathin sections of a fragment (1193-2953) containing the 12 S and 16 S ribosomal human lymphoid cell line (CEM) embedded in the gene sequences inserted into pUC19 (pMT rRNA). The second is hydrosoluble resin Lowicryl K4M. Using this cell material the Xbal 3-4 fragment (7440-8286) containing the CO II gene in enabled us to observe the good preservation of the pUC19 (pMT CO II) and the third, the Mbol fragment structure of many mitochondria, permitting us to clearly (8729-10 254) encompassing part of the ATPase-6 and CO III gene distinguish at least three components in the organelle: the in SP6 plasmid (pMT CO III) (Nelson et al. 1989). The control external envelope (outer and inner membranes), the probes used were the nuclear ribosomal genes containing cristae (inner membrane) and the matrix. This well- sequences for 18 S, 5.8 S and 28 S in pBR322 (pBR nrRNA) (Escaig-Haye et al. 1989) and the pBR322 plasmid. defined morphology was obtained after attempting several Each DNA fragment was labeled by nick translation using Bio- protocols of fixation, always including the presence of dUTP (Enzo, Biochemical), with a final length of approximately glutaraldehyde at low concentrations in a mixture with 200-400 nucleotides (Lawrence et al. 1986; Escaig-Haye et al. paraformaldehyde (see Materials and methods). We have 1989). The reaction was stopped by adding 10 mM EDTA and the observed, as previously noted by Binder et al. (1986), that DNA was precipitated overnight at —20°C. After centrifugation the use of paraformaldehyde alone for hybridization for lh, incorporated nucleotides were removed by extensive studies at the electron microscope level performed on washing in 70% ethanol. The precipitate was kept drying for biological material embedded in Lowicryl 4KM does not 2-3 h and was then suspended in hybridization buffer consisting sufficiently preserve the cellular morphology. Attempts of 4xSSC (0.6 M sodium chloride, 0.06 M sodium citrate), 50% of with other resins such as LR gold or Lowicryl HM 20 freshly deionized formamide (Fluka) (the quality of this reagent is a key in the protocol, as pointed out by Singer et al. 1987), 10 % of indicated an improvement, but were discarded because of extemporarily prepared dextran sulfate and lmgml"1 each of the induction of high levels of nonspecific labeling. The salmon sperm DNA and Escherichia coli RNA. The probe solution probes used in this study were specific for different was stored at -20°C until used. populations of mitochondrial RNA, present more or less abundantly. The first probe (pMT rRNA) contained the Hybridization sequences for ribosomal RNA, the second (pMT CO III) the sequences for ATPase-6/CO III and the third (pMT CO II) After denaturation for 3 min in boiling water, biotinylated probes those for CO II. Probes similar to the latter two were used were immediately cooled at 0°C. Hybridization was performed by

852 F. Escaig-Haye et al. t •*•

t .-.*:

Fig. 1. Ultrastructural immunogold detection of mtrRNA in CEM cells. (A) and (B) show thin section electron microscopy of CEM cells embedded in Lowicryl K4M after contact with biotinylated labeled mitoribosomal probe pMT rRNA. Hybrids are detected after indirect immunocytochemistry reaction with anti-rabbit antibiotin as a first antibody and goat anti-rabbit conjugated with gold particles of 10 nm as a secondary antibody. The labeling visualizing mtrRNA is concentrated over mitochondria profiles (empty arrows). Very few gold particles are seen over the cytoplasm containing numerous free and attached endoplasmic membrane ribosomes (arrows). N, nucleus; Nu, nucleolus; ic, interchromatin granules; V, vacuoles. (A) x 18 000 and (B) X31000. Inserts show well-characterized mitochondria exhibiting the preferential localization of mtrRNA over the inner membrane of the organelle, particularly cristae. Several clusters of gold particles in alignment or curvilinear are observed (arrowheads). (A) x 31000 and (B) x 42 000.

Ultrastructural detection of human mitochondrial RNA 853 in earlier optical studies (Mita et al. 1989). We conducted a loss of cell morphology was barely more substantial than large-scale quantitative analysis to evaluate and compare during the first hour, and subsequently remained precisely the relative abundance of each type of transcript. unchanged. So a compromise was necessary, and for In addition, among experiments to control carefully the practical reasons we chose to hybridize overnight specificity of our hybridization reaction, several were also (14-16 h) for all experiments carried out in this study. In examined at the quantitative level to strengthen the addition, we observed that the best concentration of the validity of the interpretation of the data. probe was in the range of 10 to 20^gml~1, whereas a concentration of ljugml"1 provided, in our experimental Expression of mtrRNA in lymphoblastoid cells conditions, only half of the maximum signal. Mitoribosomal RNA (mtrRNA) was localized with a 1.5 kb Xbal fragment of mitochondrial genome containing se- Control experiments quences of the ribosomal 16 S and 12 S regions (pMT We undertook several controls with various strategies to rRNA) (Nelson et al. 1989). Immunogold labeling carried analyze correctly the labeling observed over different out after hybridization of ultra-thin cell sections revealed areas of the cell and distinguish the signal as precisely as that mtrRNA visualized by gold particles was principally possible from background noise. Firstly, a control biotyni- confined to the mitochondria (Fig. 1). Quantitative analy- lated probe pBR322, containing no mitochondrial se- sis indicated that 83 % of the cellular labeling (Table 1) quences, reacted poorly with the ultrathin sections was present over mitochondria representing only 11.7 % of (Fig. 4C). The quantitative analysis indicated values of the total cell surface examined. The distribution of the 1.93±0.37, 0.91±0.19 and 1.10±0.17 for gold particle labeling densities over the mitochondria (Fig. 2) shows densities over the mitochondria, nucleus and cytoplasm, that a very large number of mitochondria contained the respectively. Secondly, we used an inappropriate probe for mtrRNA signal (95 %). The mean density of gold labeling 2 the mitochondria constituted by ribosomal DNA frag- was 52.26±2.70 gold particles,um~ of the mitochondria ments of nuclear origin (pBR nrRNA). This probe had surface. The use of 10 nm colloidal gold particles offering a previously shown that the only complementary reactive very high resolution indicated that they were not spread RNA sequences in Lowicryl cell sections were present randomly over the mitochondria. Detailed examination of principally in the cytoplasmic ribosomes and polyribo- several morphologically well-characterized mitochondria somes and in the dense fibrillar part of the nucleolus profiles, as illustrated in the insets of Fig. 1, indicated that (Escaig-Haye et al. 1989). Fig. 4A and B shows that the more than 80 % of gold particles were localized preferen- contact of the biotinylated pBR nrRNA probe with tially over the inner membrane of the organelle, particu- Lowicryl CEM cell sections yields intense gold labeling larly the cristae, demonstrating that the majority of the over the nucleolus and cytoplasm, while the mitochondria mtrRNA were closely associated with the mitochondria are apparently spared. To specify the reactivity of the membrane system. The distribution of the gold particles mitochondria surface in this hybridizing condition, we also showed that they were frequently gathered in clusters calculated the signal density over the organelle, which of 2-6 particles, either in alignment or curvilinearly. corresponds to 4.23±0.63 gold particles fim~2. It should be noted that, in some cases, certain gold particles were Time course of hybridization difficult to locate precisely, when the borderline of the The well-defined signal observed over the mitochondria organelle was not clearly defined. Thirdly, we evaluated profiles facilitated the study of one of the hybridization the level of noise introduced by each reagent of the parameters at the ultrastructural level; namely, its time detection system. By eliminating the probe from the course. Indeed, it was necessary to know at which time hybridization solution, we tested the reactivity of the first point in the saturation process all the accessible sequences antibody to be in contact with the sections, which, after were hybridized with the probe and verify whether the reacting with the colloidal gold-labeled secondary anti- state of the cell morphology was altered after prolonged body, yielded gold particle density values of 2.96±0.47, incubation in the hybridization solution. The results of the 0.74±0.15, 0.50±0.05 gold particles ,um~2, respectively for experiment are illustrated by the kinetic curve (Fig. 3), the mitochondria, nucleus and cytoplasm (Fig. 4D, which shows that the hybridization reaction reached a Table 2). In the absence of the primary antibody, non- peak after 9 h incubation, then stabilized (or very slightly specific binding with the gold-labeled secondary antibody decreased at 16 h). At the time of maximum reaction the alone was extremely low (—1.1 gold particles fan~2) for

Table 1. Total surface and total number of gold particles for each cell compartment, according to the probe used Surface (fan2) No. of particles Probes pMT rRNA pMT CO m pMT CO H pBR322 pMT rRNA pMT CO m pMT CO II pBR322 Total 274.48 286.55 236.35 370.54 1844 603 575 415 Nucleus 71.14 118.62 129.05 179.92 161 171 375 189 Cytoplasm 171.15 141.09 87.11 154.89 145 132 88 151 Mitochondria 32.19 26.83 20.19 35.73 1538 300 112 75 Mean mito. 0.36* 0.42* 0.33* 0.33* 17.47t 4.76t 1.84t 0.78t % Mito. 11.7| 9.3* 8.5* 9.6$ 83§ 49.7§ 19.4§ 18§ NbH 88 63 61 97 88 63 61 97 For each probe used, the surface of mitochondria, surrounding cytoplasm and nucleus were measured by an image-analyzing system (Biocom 200) and the gold particles were counted manually over each cell compartment, on electron micrograph negatives. • Mean surface of mitochondria. t Mean number of gold particles per mitochondrion. I Percentage of surface examined which is occupied by mitochondria. § Percentage of labeling observed over mitochondria. H Number of mitochondria examined for each probe.

854 F. Escaig-Haye et al. 70

50 Fig. 2. Histogram of labeling densities. •o 45 The distribution of labeling densities § (number of gold particles per /an2 of •s 40 mitochondria) obtained with each type of mitochondrial probe, pMT rRNA (A), pMT 35 D CO III (B) and pMT CO II (C), is 30 compared to that obtained with the non- hybridizing pBR322 control probe (D). For a given density value, the percentage of § 25 Hi mitochondria represents all mitochondria fr with density values between this exact 20 value and the preceding one (e.g. density 10 represents all densities ranging from 5.1 to 10.0 inclusive). The mean density value and standard deviation of gold labeling for each probe are 52.26±2.70 (pMT rRNA), 12.25±1.81 (pMT CO III), 6.15±0.93 (pMT CO U) and 1.93±0.37 J i. . i I . . (pBR322) and the percentage of u T mitochondria with gold signals is 98, 81, 20 40 60 100 120 78 and 34 % respectively. Density each cell compartment (data not shown). These exper- protocol, was performed using RNase H treatment after iments indicate that the noise labeling is essentially the hybridization step (Fig. 4F). Using this protocol, only linked with the reactivity of the antibiotin antibody on the the cellular RNA sequences that had established hydrogen Lowicryl cell sections, rather than with false hybridiz- bound to the complementary sequences present in the ations or nonspecific sticking of the probe. Finally, in DNA probe were removed. The fact that the residual another set of experiments, we demonstrated that the labeling observed with the mitoribosomal probe fell from interaction between mitochondrial DNA probe and Lowic- 52.26±2.70 to 7.60±1.30 gold particles per unit area of ryl cell sections gives rise effectively to the formation of mitochondria provides supporting evidence for the prior RNA-DNA hybrids, since ultrathin sections from which presence of ribosomal RNA sequences in the organelle. cellular RNA was removed by RNase A treatment The RNase H treatment could be performed either before (lmgml""1 in 2xSCC for lh) before hybridization showed or after the immunocytochemistry stage, but in the latter labeling similar to that observed with the control pBR case, we observed that the enzymatic digestion was much probe (Fig. 4E). In addition, a more rigorous analysis, as less efficient. suggested by Lawrence et al. (1989) and following their Detection of mitochondrial messenger RNA We next analyzed different populations of mitochondrial messenger RNA (mRNA) with pMT CO III probe coding for the ATPase-6 and CO III, and with pMT CO II probe coding for CO II (Nelson et al. 1989) (Fig. 5A and B). With the first probe, 41 % of mitochondria contained the double information for ATPase-6 and CO II (Fig. 5A). The density of the overall labeling corresponding to the accumulation of these two distinct mRNA populations was 12.25±1.81 gold particles/on"2, representing 49.7% of the total labeling concentrated over only 9.3% of the cellular area examined (Table 1). After hybridization of ultrathin sections with the second probe pMT CO II (Fig. 5B), the 16 indirect immunocytochemistry system of detection re- Time in hours vealed the presence of hybrids visualized by gold particles over 43 % of mitochondria profiles (Fig. 2). The mean Fig. 3. Time course of a hybridization reaction on thin density of labeling for the CO II message was 6.15±0.93 Lowicryl-embedded cell sections incubated with the gold particles fim~2, representing 19.4% of the cellular biotinylated labeled mitoribosomal probe. Points represent an labeling over 8.5 % of the cell surface analyzed (Table 1). average of two experiments, each performed on duplicate grids. When the distribution of gold particles over the mitochon- Data were plotted as a percentage of the maximum dria was further analyzed, it was found that, with both hybridization value obtained after 9 h, and corresponding to a mean density of 48 gold particles fan , calculated by dividing these probes, labeling was predominantly associated with the total number of gold particles by the total mitochondria the cristae membrane (80%), suggesting proximity be- surface). tween mRNA molecules and the mitochondrial membrane Ultrastructural detection of human mitochondrial RNA 855 ft • c•

Nu •/?•

<>•• V

•—m •A

Fig. 4. Ultrastructural immunogold detection in control experiments using unrelated mitochondrial probes (A, B and D), hybridization buffer (C), and RNase H and RNase A enzymatic digestion (E and F). (A and B) Nuclear ribosomal probe reacts essentially with the nucleolus and cytoplasmic ribosomal structures, while mitochondria are apparently spared (empty arrows). The insert shows labeling in close contact with the external envelope (arrows). (A,B) 18000; inset, X23 000. (C) Background level of immunolabeling obtained with hybridization buffer, x 20 000. (D) Background level of immunolabeling obtained with the non- hybridizing pBR322 probe, x 23 000. (E) Partially reduced labeling observed after RNase H treatment performed after hybridization with the pMT rRNA probe, x 16 000. (F) Labeling reduced to background level observed after RNase A treatment performed before hybridization with the pMT rRNA probe, x 16 000. N, nucleus; Nu, nucleolus; V, vacuole; ic, interchromatin granules; m, mitochondria free of labeling. Gold particles in mitochondria are indicated by arrowheads.

856 F. Escaig-Haye et al. Table 2. Density of labeling Hybridization Probes pMT rRNA pMT CO III pMT CO D pBR322 buffer RNaseH Mitochondria 52.26±2.70 12.25±1.81 6.15±0.93 1.93±0.37 2 96±0.47 7.60±1.30 Nucleus 2.20±0.35 1.62±0.21 2.87±0.36 0.91±0.19 0.74±0.15 1.18±0.24 Cytoplasm 0.71±0.12 0.92±0.14 1.07±0.13 1.10±0.17 0.50±0.05 0.40±0.02 P* <0.01 <0.05 <0.001 _ _ _

Values are numbers of gold particles /im 2 of the cell compartment surface±standard deviation. * P value determined by Student's <-test. When the nuclear labeling density yielded by control pBR322 is compared with that yielded by pMT rRNA, pMT CO III and pMT CO II, respectively, the results are significant in each case, indicating that the nucleus reacts with the mitochondrial probes. system. In contrast to mtrRNA, however, very rare for hybridization signals and an efficient, low-temperature clusters of colloidal gold particles were seen. cell-embedding procedure in the polar Lowicryl K4M resin, which is well known to preserve the structure and Extramitochondrial reactivity of mitochondrial probes protein antigenicity of the mitochondria (Bendayan and The different protocols performed to ascertain the speci- Shore, 1982; Bendayan et al. 1983; Martinez-Ramon et al. ficity of our hybridization conditions permitted us to 1990). Though the optimal conditions of fixation with a compare the density of gold particles for each population of higher concentration of glutaraldehyde were not transcripts over the different cell compartments. As shown employed, because that would decrease the efficiency of in Table 2, we can observe that in addition to the the hybridization reaction (Binder et al. 1986; Trembleau mitochondria surface where the appropriate gene ex- et al. 1988), the preservation of the general cell ultrastruc- pression occurs, the hybridizing mitochondrial probes ture was acceptable even after prolonged incubation in the seemed to react also with the nucleus, on which the gold hybridization buffer solution (Binder et al. 1986; Escaig- particle densities were 2.87±0.36, 2.20±0.35 and Haye et al. 1989). In the present case, many mitochondria 1.62±0.21 gold particles [an'2 for CO II, rRNA and of CEM cells continued to exhibit their membrane ATPase-6/CO III, respectively. These nuclear values for network included in the matrix space. We have estab- each probe used were significantly higher than those lished here that the optimal hybridization time was at observed in their cytoplasmic counterparts. Moreover, least 9h, with 80% obtained after 5h. These results, when the nuclear labeling obtained with the mitochon- described for the first time at the ultrastructural level, drial probes was compared with that observed in control indicate a slightly longer time than previously found experiments performed with the hybridization buffer (Binder et al. 1986). alone (0.74±0.15) or containing the pBR322 nonhybridiz- Several lines of evidence indicate that the target ing probe (0.91 ±0.19), statistical calculations based on molecules that we visualized with the biotinylated Student's i-test also revealed a significant signal over the mitochondrial probes correspond effectively to the tran- nucleus. scripts of the appropriate human mitochondrial genes. Firstly, for each class of RNA, whether ribosomal or Detection of ribosomal RNA transcript in normal human messenger, the quantitative data obtained by gold particle cells and tissues counts after immunocytochemistry indicate that the In order to extend our protocol to in vivo studies, several mitochondria compartment represents the main subcell- experiments were carried out with human tissues embed- ular region in which the molecules were found. Secondly, ded in Lowicryl K4M resin. The ultrathin sections from an the labeling intensity was highest for rRNA, then ATPase- appendix removed surgically, lymph-node and muscle 6/CO III and, lastly, CO II, which correlates well with the biopsies, and cultured peripheral blood mononuclear cells quantitative molecular data on the relative proportions of were tested with the mitochondrial probe specific for mitochondrial RNA present in cultured human cells mtrRNA. As shown in Fig. 6, specific hybridization as (Attardi, 1987). Finally, the hybridization response was visualized by numerous gold particles was restricted to sensitive to RNase A and RNase H, and a lower mitochondria, for each tissue sample examined. It is background signal was obtained after contact of the thin noteworthy that the protocol of embedding could be sections with the hybridization solution, either alone or delayed after fixation, by keeping the sample in buffered containing the non-hybridizing pBR322 probe. As far as the labeling yielded by the nuclear ribosomal gene probe is sucrose solution. In these conditions, there was no 2 substantial loss of hybridization labeling or morphology. concerned, though lower (4.23±0.63 gold particles/xm~ ), the difference was not statistically different from that obtained with the less efficient mitochondrial probe (pMT CO II) (6.15±0.93), but was statistically different from the Discussion unrelated pBR322 probe (1.93+0.37). This suggests that The present study demonstrates for the first time the the nuclear ribosomal probe does not introduce a high visualization of human mitochondrial transcripts in background, but rather that possible information con- relation to the ultrastructural morphology of the organ- tained in it hybridizes with certain RNA sequences elle. In particular, our results demonstrate that mitochon- present in the mitochondria. This explanation seems drial RNAs of the ribosomal or messenger type have an plausible, since it has been demonstrated that the intra-organelle location, preferentially on or close to the replication of the human mitochondrial DNA needs the inner membrane system, the relationship of this location presence of the specific enzyme mtDNA primase, which, to with the process of mitochondrial protein synthesis will be be active, must be associated with small RNAs, one of discussed below. These observations were made possible which has been identified as the nuclear 5.8 S rRNA by the joint use of a highly resolutive colloidal gold marker (Wong and Clayton, 1986). Alternatively, a fraction of the

Ultrastructural detection of human mitochondrial RNA 857 - • r.

Fig. 5. Ultrastructural immunogold detection of mitochondrial mRNA in CEM cells. (A) Thin section electron micrograph of CEM cells hybridized with the pMT CO III probe shows transcripts corresponding to CO III and ATPase 6 (arrowheads) in mitochondria. Nu, nucleolus. A cluster of gold particles is seen near the external envelope (arrow) indicated by two unit membranes (double arrows). X31000. (B) With the pMT CO II probe, visualization of messenger for CO II in the organelle (arrowheads). A group of unlabeled mitochondria is indicated by a star, x 23 000. The inset shows at high magnification that the gold signal observed over this particularly well labeled mitochondria is preferentially associated with the inner mitochondrial membrane. A cluster of gold particles is visible (arrowhead). X40 000. Gold particles located outside the mitochondria are indicated by circles. mitochondrial labeling observed could be linked to the specific probe for only the 5.8 S rRNA, to clarify at the sub- hybridization of some cytoplasmic ribosomes bound to the cellular level the presence of this surprising nuclear- surface of the outer membrane (Kellems et al. 1974). encoded RNA inside the mitochondria (Wong and Clayton, Further extensive analysis would be necessary with a 1986; Chang and Clayton, 1987).

858 F. Escaig-Haye et al. \

•-'••

N .•* B

V I N f

Fig. 6. Ultrastructural immunogold detection of mtrRNA in human tissues. Thin section electron microscopy of human cultured PBMC (A), lymph node (B), appendix (C) and muscle (D) embedded in Lowicryl K4M were hybridized with mitoribosomal probe, showing mitochondria-localized labeling (arrows). N, nucleus. (A) x 21000, (B) x 42 000, (C) x 28 000 and (D) x 70 000. In (D), mitochondria are observed between longitudinal sections of myofibrils (f). Insert shows mitochondria in a muscle connective tissue cell, x 30 000.

The qualitative and quantitative immunogold labeling Attardi, 1981), because of a factor involved in the analysis of mitochondria transcripts indicates that the termination of a transcription of rRNA genes on the H most prominent RNA species within the organelle is the strand DNA (Christianson and Clayton, 1986; Kruse et al. rRNA. This could be explained by the fact that, in human 1989); (ii) the half-life of rRNA is longer than that of cells: (i) the transcription rate of the ribosome genes is 15- mRNA (Gelfand and Attardi, 1981). The association of the to 60-fold higher than that of the other genes (Gelfand and rRNA with proteins leads to the formation of mitochon- Ultrastructural detection of human mitochondrial RNA 859 drial ribosomes (mitoribosomes), which can be observed in hybridization between certain cellular RNA sequences situ by conventional electron microscope studies in various and the mitochondrial probes. Recently, mitochondrial species and in cultured human cells (Munn, 1974; Curgy, RNA sequences have been detected in the nucleus of 1985). In our case, direct visualization of mitoribosomes is, Drosophila cells (Tourmente et al. 1990). We cannot however, not possible because of the inappropriate exclude this possibility in the human cells used in this embedding procedure. Thus, the only means of identifying study where it is known that mitochondrial DNA these intraorganelle components is their visualization by sequences are present in the nucleus (Tsuzuki et al. 1983; our gold-labeling methodology, which results from hybrid- Kristensen and Prydz, 1986; Kamimura et al. 1989). ization to their rRNAs. The detailed examination of the Though a direct approach for evaluating the sensitivity gold particle distribution over the mitochondria sections of our methodology is not possible, we can propose an indicates that the majority (—80 %) of mitoribosomes are approximate correlation between the gold signal observed in close contact with the inner membrane system of the on sections and the number of molecules estimated to be mitochondria. This is consistent with a body of morpho- present in the mitochondria. When we based our calcu- logical work, showing that the mitoribosomes are prefer- lation on the CO II mRNA, an average signal close to one entially located near the cristae or the inner membrane gold particle (excluding background noise) was observed (Swift, 1965; Vignais et al. 1969, 1972; Kuriyama and per mitochondria profile area (mean value 0.33/an2) Luck, 1973). In addition, the frequent observation of gold (Table 1). After morphometric analysis indicating a mean clusters in alignment or curvilinearly suggests that number (n) of 4xlO~2 mitochondria per /an2 of cytoplas- mitoribosomes could be associated in polysome structures. mic surface and that a percentage (p) of 1 % of the Such structures have been observed in situ and after cytoplasmic surface was occupied by the mitochondria, we isolation (Vignais et al. 1969). Moreover, in HeLa cells, at applied the formula: N=n^'2/1.382 p1/2 (Weibel and least 50 % of mitoribosomes can be recovered in the form of Gomez, 1962) to obtain a numerical density (iV) of 0.05 polysomes (Ojala and Attardi, 1972). mitochondria per /an3. Knowing that the mean cytoplasm The nature of the mitochondrial rRNA distribution volume in our cells was evaluated at 2x 103 fan3, we found observed provides fundamental insights into the basic a mean number of approximately 100 mitochondria per mechanism of mitochondrial protein synthesis. The mem- cell. From this, and quantitative molecular data indi- brane location of the ribosome structures, which constitute cating that, in human cultured cells, the order of an essential element in the apparatus of protein synthesis, magnitude of the copy number is roughly 1200 for CO II supports the hypothesis of a membrane site for protein mRNA, we can estimate that there are 12 copies of CO II synthesis. The preferential accumulation of mRNA mRNA (Attardi, 1987) molecules per mitochondria. Conse- detected also in the proximity of the inner membrane quently, it would appear that, in the present detection surface strengthens this assumption. Biochemical exper- system, a specific signal of one gold particle on the iments have previously provided some evidence that mitochondria surface after sectioning corresponds to the mitochondrial protein synthesis requires the presence of presence of 12 RNA molecules in the mitochondrial an intact mitochondrial inner membrane (Munn, 1974). volume. Analyzing the influence of temperature on mitochondrial The approach used in this report may contribute protein synthesis, Towers et al. (1973) concluded that the significantly to the analysis of human mitochondrial DNA ribosomes have an intimate functional relationship with expression, complementing blot analysis and in situ the inner mitochondrial membrane. The implication of the detection at the cellular level (Mita et al. 1989; Shoubridge membrane system may be linked to the membrane nature et al. 1990; Moraes et al. 1991). Given the possibility it of the mitochondrial DNA-encoded proteins, in the same offers of visualizing the fine structure of the mitochondria, way that the protein synthesis of the cell membrane it may facilitate studies on the correlation between protein occurs on the ergastoplasm membrane structure. mitochondrial genome activity and morphological changes It seems also that the membrane system could be in the organelle. Various structural modifications to the physically associated with mitochondrial DNA (Munn, mitochondria have been observed in different physiologi- 1974; Albring et al. 1977). In the course of further studies, cal and pharmacological states (Munn, 1974; Shea et al. we shall adapt a protocol for specific detection of 1990), in the ageing process (Trounce et al. 1989; Martinez- mitochondrial DNA, which would make it possible to Ramon et al. 1990) and clinical disorders (Munn, 1974). On examine this hypothesis. the latter point, growing interest has emerged since the Through RNA molecule detection, we have noted above discovery of alterations in the mitochondrial genome that the signal density yielded by each mitochondrial associated with several human mitochondrial cytopathies, probe over mitochondria, was significantly different from involving particularly neuromuscular diseases (Holt et al. that observed over the cytoplasm and nucleus, demon- 1988; Lestienne and Ponsot, 1988; Zeviani et al. 1988; strating that the mitochondria constitute the main cell Rotig et al. 1988; Wallace et al. 1988; Shoffner et al. 1990). area where the RNA molecules are concentrated. Analysis These alterations have generally been found to be of the data indicates also that, for the mitochondrial heteroplasmic and multisystemic (Nelson et al. 1989) and probes, the nuclear signal was significantly higher than the physiopathological mechanism determining which the cytoplasmic one. This was unexpected and could be organ is specifically affected seems to depend on the interpreted in terms of higher background induced by balance between altered and normal mitochondrial DNA some particular nuclear components that react with these (Wallace, 1989; Nakase et al. 1990). Recent reports have probes. However, the absence of specific labeling after demonstrated that the deleted mitochondrial genome RNase treatments and the significantly lower nuclear remains transcriptionally active (Mita et al. 1989; Shou- colloidal gold signal yielded by the hybridization solution bridge et al. 1990; Moraes et al. 1991). With the approach alone or containing the non-hybridizing pBR322 probe, used in our study, we can investigate the proportion of compared to that obtained with each mitochondrial probe mitochondria expressing (or not) certain genes, and (see Table 2), indicate that the light labeling observed over explore the possibility of intraorganelle complementation the nucleus would be more likely to be the result of cross- between mutant and wild-type DNA. In the present paper, 860 F. Escaig-Haye et al. we have shown that this research can be conducted not dans des cellules de hamster transformers par le virus de Rous. C.R. Acad. Sci. Paris, sSrie D, 1668-1670. only in isolated cells, such as lymphocytes, but also with GELFAND, R. AND ATTARDI, G. (1981). Synthesis and turnover of human tissues, particularly muscle tissue. mitochondrial ribonucleic acid in HeLa cells: the mature ribosomal Finally, several reports have indicated the presence of and messenger ribonucleic acid species are metabolically unstable. viral materials inside mammalian mitochondria (Gazzolo Molec. cell. Biol. 1, 497-511. et al. 1969; Lunger and Clark, 1976). We think that the GRIGORIEV, V. B., KADOCHNICOV, J. P., RUDNEVA, A., KARAMOV, E. V., KLIMENKO, S., ESCAIG, F. AND FOURNIER, J. G. (1989). Detection by approach presented here could usefully be applied in this immuno gold techniques of HIV antigenes in Lowicryl ultrathin field, to gain insight into mitochondria-virus interactions sections of infected cells. Path. Res. Pract. 184, 494-497. in human cells. It would also be of interest to clarify the HOLT, I. J., HARDING, A. E. AND MORGAN-HUGHES, J. A. (1988). involvement of the mitochondria in human scrapie-like Deletions of muscle mitochondrial DNA in patients with diseases such as Creutzfeldt-Jakob and Kuru (Aiken et al. mitochondrial myopathies. Nature 331, 717-719. KAMIMURA, N., ISHII, S., LIANDONG, M. AND SHAY, W. (1989). Three 1990; Aiken and Marsh, 1990). separate mitochondnal DNA sequences are contiguous in human genomic DNA. J. molec. Biol. 210, 703-707. The authors are greatly indebted to Dr C. Marsac and Dr M. KELLEMS, R E., ALLISON, V. F. AND BUTOW, R. A. (1974). Cytoplasmic Fardeau for muscle biopsy samples, Dr S. Pr6vot for appendix and type 80 S ribosomes associated with yeast mitochondria. Evidence for lymph node samples, and to D. Touret for illustrations. This work the association of cytoplasmic ribosomes with the outer mitochondrial was supported by the Association Franchise Contre la Myopathie. membrane in situ. J. biol. Chem. 10, 3297-3303. KRISTENSEN, T. AND PRYDZ, H. (1986). The presence of intact mitochondrial DNA in HeLa cell nuclei. Nucl. Acids Res. 14, 2597-2609. References KRUSE, B , NARASIMHAN, N. AND ATTARDI, G (1989). Termination of transcription in human mitochondria: identification and purification AIKEN, J. M. AND MARSH, R. F. (1990). The search for acrapie agent of a DNA binding protein factor that promotes termination. Cell 58, nucleic acid. Microbiol. Rev. 54, 242-246. 391-397. AIKEN, J. M., WILLIAMSON, J. L., BORCHARDT, L. M. AND MARSH, R. F. KURIYAMA, Y. AND LUGK, D. J. L. (1973). Membrane-associated (1990). Presence of mitochondrial D-loop DNA in scrapie-infected ribosomes in mitochondria of Neurospora crassa. J. Cell Bwl. 59, brain preparations enriched for the prion protein. J. Virol. 64, 776-784. 3265-3268. LANGER, P. R., WALDROP, A. A. AND WARD, D. C. (1981). Enzymatic ALBRING, M., GRIFFITH, J. AND ATTARDI, G. (1977). Association of a synthesis of biotin-labeled polynucleotide novel nucleic acid probes. protein structure of probable membrane derivation with HeLa cell Proc. natn Acad. Sci. U.S.A. 78, 6633-6637 mitochondrial DNA near its origin of replication. Proc. natn. Acad. LAWRENCE, J. B., SINGER, R. H. AND MARSELLE, L. M. (1989). Highly Sci. U.S.A. 74, 1348-1352. localized tracks of specific transcripts within interphase nuclei ANDERSON, S., BANKIER, A. T., BARRELL, B. G., DE BRUIJN, M. H. L., visualized by in situ hybridization. Cell 57, 493-502. COULSON, A. R., DROUIN, J., EPERON, I. C, NIERLICH, D P., ROE, B. LESTIENNE, P. AND PONSOT, G. (1988). Kearns-Sayre syndrome with A., SANGER, F., SCHREIER, P. H., SMITH, A. J. H., STADEN, R. AND muscle mitochondrial DNA deletion. Lancet 1, 885. YOUNG, I. G. (1981). Sequence and organization of the human LUNGER, P. D. AND CLARK, H. F. (1976). An ultrastructural study and raitochondrial genome. Nature 290, 457-465. model of membrane—intramitrochondrial virus relationships. ATTARDI, G. (1987). The elucidation of the human mitochondrial Intervirology 7, 240-249. genome. In Advances in Myochemistry (ed G. Benzi), vol. 1, pp. MARTINEZ-RAMON, A., KNECHT, E., RUBIO, V. AND GRISOLIA, S. (1990). 157-171. John Libbey Eurotext Ltd. Levels of carbamoyl phosphate synthetase I in livers of young and old BENDAYAN, M. (1984). Protein A gold electron microscopy rats assessed by activity and immunoassays and by electron immunocytochemistry: methods, applications and limitations. J. microscopic immunogold procedures. J. Histochem. Cytochem. 38, Electron Microsc. Tech. 1, 243-270. 371-376. BENDAYAN, M., REDDY, M. K., HASHIMOTO, T. AND REDDY, J. K (1983). MITA, S., SCHMIDT, B., SCHON, E A., DIMAURO, S. AND BONILLA, E. Immunocytochemical localization of fatty acid metabolizing heat- (1989). Detection of 'deleted' mitochondrial genomes in cytochrome-c- stable and heat-labile enoyl-coenzyme A (CoA) hydratases in liver and oxidase-deficient muscle fibers of a patient with Kearn-Sayre renal cortex. J. Histochem. Cytochem 31, 509-516. syndrome. Proc. natn Acad. Sci. U.S.A. 86, 9509-9513 BENDAYAN, M. AND SHORE, G. C. (1982). Immunocytochemical MORAES, C. T., ANDREETTA, F., BONILLA, E., SHANSKE, S., DIMAURO, S. localization of mitochondrial proteins in the rat hepatocyte. J. AND SCHON, E. (1991). Replication-competent human mitochondrial Histochem. Cytochem. 30, 139-147. DNA lacking the heavy-strand promoter region. Molec. cell. Bwl. 11, BINDER, M., TOURMENTE, S., ROTH, J., REBAUD, M. AND GBHRING, W. J. 1631-1637 (1986). In situ hybridization at the electron microscope level: localization of transcripts on ultrathin sections of Lowicryl K4M- MUNN, E. A. (1974). The Structure of Mitochondria. Academic Press: embedded tissue using biotinylated probes and protein A-gold London. complexes. J. Cell Bwl. 102, 1646-1653. MURPHY, W., ATTARDI, B., TU, C. AND ATTARDI, G. (1975). Evidence for complete symmetrical transcription in vivo of mitochondrial DNA in BOGENHAGEN, D. F., ApPLEGATE, E. F. AND YoZA, B. K. (1984). Identification of a promoter for transcription of the heavy strand of HeLa cells. J. molec. Biol. 99, 809-814. human mtDNA: in vitro transcription and deletion mutagenesis. Cell NAKASE, H , MORAES, C, RIZZUTO, R., LOMBES, A , Di MAURO, S. AND 36, 1105-1113. SCHON, E. (1990). Transcription and translation of deleted BRANGEON, J., NATO, N. AND FOROHIONI, A. (1989). Ultrastructural mitochondrial genomes in Kearns-Sayre syndrome: implications for detection of ribulose-l,5-biphosphate carboxylase protein and its pathogenesis. Am. J. hum. Genet. 46, 418-427. subunit mRNAs in wild type and holoenzyme-deficient Nicotiana NELSON, I., PONSOT, G , MARSAC, C, DEGOUL, F., VAYSSIERE, J. L., using immuno-gold and in situ hybridization techniques. Planta 177, OBERMAIER, B., ROMERO, N., FARDEAU, M. AND LESTIENNE, P. (1989). 151-159. General heteroplasmic mitochondrial DNA markers in Kearns-Sayre CHANG, D. D. AND CLAYTON, D. A. (1984). Precise identification of syndrome. Nucl Acids Res. 17, 8117-8127. individual promoters for transcription of each strand of human OJALA, D. AND ATTARDI, D. (1972). Expression of the mitochondrial mitochondrial DNA. Cell 36, 635-643. genome in Hela cells Properties of mitochondrial polysome. J. molec. CHANG, D. D. AND CLAYTON, D. A. (1987). A mammalian mitochondrial Biol. 65, 273-281. RNA processing active nucleus-encoded RNA. Science 235, 1178-1184. OJALA, D., MONTOYA, O. D. AND ATTARDI, G. (1981). The tRNA CHRISTIANSON, T. W. AND CLAYTON, D. A. (1986). In vitro transcription punctuation model of RNA processing in human mitochondria. Nature of human mitochondrial DNA: accurate termination requires a region 290, 470-474. of DNA sequence that can function bidirectionally. Proc. natn. Acad. RADIC, M. Z., LUNDGREN, G. AND HAMKALO, B. A. (1987). Curvature of Sci. U.S.A. 83, 6277-6281. mouse satellite DNA and condensation of heterochromatin. Cell 50, CURGY, J. (1985). The mitoribosomes. Biol. Cell 54, 1-38. 1101-1108. ESCAIG-HAYE, F., GRIGORIEV, V. AND FOURNIER, J. G. (1989). Detection ROTIG, G., COLONNA, M., BONNEFONT, J., BLANCHE, S., FISHER, A., ultrastructurale d'ARN ribosomal par hybridation in situ a l'aide SAUBUBRAY, S. AND MUNNICH, A. (1988). Deletion of blood d'une sonde biotinyl^e sur coupes ultrafines de cellules animales en mitochondrial DNA in pancytopenia. Lancet 2, 567-568. culture. C.R. Acad. Sci. Paris, sdrie III, 429-434. SHEA, C. R., SHERWOOD, M. E., FLOTTE, T J., CHEN, N., SCHOLZ, M. AND GAZZOLO, L., DE-THE, G., VIGIER, P. AND SARMA, P. (1969). Presence de HASAN, T. (1990) Rhodamine 123 phototoxycity in laser-irradiated particules a l'aspect de nucle'ocapsides associe'es aux mitochondries MGH-U1 human carcinoma cells studied in vitro by electron

Ultrastructural detection of human mitochondrial RNA 861 microscopy and confocal laser scanning microscopy. Cancer Res. 50, (1973). The biogenesis of mitochondnal 29. Effects of temperature- 4167-4172 induced phase changes in membranes of protein synthesis by SHOFFNER, J. M., LOTT, M. T., LEZZE, A. M S., SEIBEL, P., BALUNGER, S. mitochondria. Biochim. biophys. Ada 299, 153-161. W. AND WALLACE, D. C. (1990). Myoclonic epilepsy and ragged red TREMBLEAU, A., FEVRE-MONTANGE, M. AND CALAS, A. (1988). fiber disease (MERRF) is associated with a mitochondrial DNA Localisation ultrastructurale d'ARNm codant pour l'ocytocine par tRNAlyB mutation. Cell 61, 931-937. hybridation in situ. Etude par radioautographie a haute resolution a SHOUBRIDGE, E. A., KARPATI, G. AND HASTINGS, K E M. (1990). l'aide d'une sonde oligonucl^otidique tritie'e. C.r. Acad. Sci. Paris, 307, Deletion mutants are functionally dominant over wild-type sirie III, 869-874 mitochondrial genomes in skeletal muscle fiber segments in TROUNCE, I., BYRNE, E. AND MAHZUKI, S. (1989). Decline in human mitochondrial disease. Cell 62, 43-49. skeletal muscle mitochondrial respiratory chain function, possible SHUEY, D. AND ATTARDI, G. (1985). Characterization of an RNA factor in ageing. Lancet 1, 637-639. polymerase activity from HeLa cell mitochondria, which initiates TSUZUKI, T., NOMIYAMA, H , SETOYAMA, C, MAEDA, S. AND SHIMADA, K. transcription at the heavy strand rRNA promoter and the light strand (1983). Presence of mitochondrial DNA-like sequences in the human promoter in human mitochondrial DNA J. biol. Chem. 260, nuclear DNA. Gene 25, 223-229. 1952-1958. VIGNAIS, P. V , HUET, J. AND ANDRE, J. (1969). Isolation and SILVA, F. G., LAWRENCE, J. B. AND SINGER, R H. (1989). Progress characterization of ribosomes from yeast mitochondria. FEBS Lett. 3, toward ultrastructural identification of individual mRNAs in thin 177-181 section, myosin heavy chain in developing myotubes. In Techniques in VIGNAIS, P. V., STEVENS, B. J., HUET, J. AND ANDRE, J. (1972). Immunocytochemistry (ed. G. Bullock and P. Petrusz), vol. 4, pp. Mitoribosomes from Candida utilis. Morphological, physical and 77-83. London: Academic Press. chemical characterization of the monomer form and of its subunits. J. SINGER, R. H. AND WARD, D. C. (1982). Actin gene expression visualized Cell Biol. 54, 468-492. in chicken muscle tissue culture by using in situ hybridization with a WALLACE, D C , SINGH, M. T., LOTT, J. A., HODGE, T. G., SCHOURR, A. biotinated nucleotide analogue. Proc. natn. Acad. Sci. U.S.A. 79, M., LEZZA, L. Z. AND NIKOSKELAINEN, E. (1988). Mitochondrial DNA 7331-7335. mutation associated with Leber's hereditary optic neuropathy. Science SINGER, R. H., LAWRENCE, J. B. AND RASHTCHIAN, R. N (1987). Toward 242, 1427-1430. a rapid and sensitive in situ hybridization methodology using isotopic WALLACE, D. C. (1989). Mitochondrial DNA mutations and and nonisotopic probes. In In situ Hybridization: Application to the neuromuscular disease. Trends Genet. 5, 9-13. Central Nervous System (ed. K. Valentino, J. Eberwine and J. WEBSTER, H. E., LAMPERTH, L., FAVILLA, J. T., LEMKE, G., TESIN, D. AND Barchas), pp. 71-96. New York: Oxford University Press. MANUELIDIS, L. (1987). Use of a biotinylated probe and in situ SINGER, R. H., LANGEVIN, G. L. AND LAWRENCE, J. B (1989a). hybridization for light and electron microscopic localization of po Ultrastructural visualization of cytoskeletal messenger RNAs and mRNA in myelin-forming Schwann Cells. Histochemistry 86, 441-444. their associated protein using double label in situ hybridization. J. WEIBEL, E. R. AND GOMEZ, D. M. (1962). A principle for counting tissue Cell Biol. 108, 2343-2353. structures on random sections. J. appl. Physiol. 17, 343-358. SINGER, R. H., LAWRENCE, J. B., SILVA, F., LANGEVIN, G. L., POMEROY, WOLBER, R. A., BEALS, T. F. AND MAASSAB, H. F. (1989). Ultrastructural M. AND BILLINGS-GAGLIARDI, S. (19896). Strategies for ultrastructural localization of Herpes simplex virus RNA by in situ hybridization. J. visualization of biotinated probe hybridized to messenger RNA in situ. Histochem. Cytochem. 1, 97-104. Curr. Top. Microbiol. Immunol. 143, 57-69. WONG, T. W. AND CLAYTON, D. A. (1986). DNA primase of human SWIFT, H. (1965). Nucleic acids of mitochondria and chloroplasts. Am. mitochondria is associated with structural RNA that is essential for Nat. 99, 201-227. enzymatic activity. Cell 45, 817-825. THIRY, M. AND THIRY-BLAISE, L. (1989). In situ hybridization at the ZEVIANI, M., MORAES, C. T., DIMAURO, S., NAKASE, H., BONILLA, E., electron microscope level. An improved method for precise localization SCHON, E. A. AND ROWLAND, L. P. (1988). Deletion of mitochondrial of ribosomal DNA and RNA. Eur. J. Cell Biol. 50, 235-243. DNA in Kearns-Sayre syndrome. Neurology 38, 1339-1346. TOURMENTE, S., SAVRE-TRAIN, I., BERTHIER, F. AND RENAUD, M. (1990). Expression of six mitochondrial genes during Drosophda oogenesis: analysis by in situ hybridization. Cell Differ. Dev. 31, 137-149. TOWERS, N R., KELLERMAN, G. M., RAISON, J. K. AND LINNANE, A W. (Received 29 May 1991 - Accepted 28 August 1991)

862 F. Escaig-Haye et al.