Adaptation of a non-radioactive in situ hybridization method to electron microscopy: detection of tenascin mRNAs in mouse cerebellum with digoxigenin-labelled probes and gold-labelled antibodies

In this study we describe a method for the detection of mRNAs at the ultrastructural level using a non-radioactive in situ hybridization method based on digoxigenin-labelled cRNA probes and gold-labelled digoxigenin-specific antibodies. We applied this protocol to an analysis of the expression of the extracellular matrix protein tenascin in the developing cerebellar cortex of the mouse. To gain an impression of the sensitivity attainable with digoxigenin-labelled probes, we first established at the light microscopic level that the hybridization signal obtained with the non-radioactive probe is as sensitive as that obtained with a ³⁵S-labelled probe. The non-radioactive hybridization protocol was then combined with electron microscopic post-embedding and immunogold detection techniques. Tenascinspecific, digoxigenin-labelled cRNA probes were hybridized to ultrathin sections of Lowicryl K4M-embedded tissue and the probe/target mRNA hybrids were detected using gold-labelled antibodies to digoxigenin. In agreement with the observations from in situ hybridization at the light microscopic level, specific labelling was observed in Golgi epithelial cells in the region of the Purkinje cell layer and cells in the internal granular layer, which could be identified as astrocytes by ultrastructural criteria. Labelling was detectable in association with free ribosomes and ribosomes of the rough endoplasmic reticulum. In addition, focal hybridization signals were occasionally found in the nucleus. No signal was observed in Golgi epithelial cells or astrocytes using sense or in any other cerebellar cell type using either sense or anti-sense probes. The described in situ hybridization technique uses ultrastructural criteria to associate the presence of a given mRNA species with a particular cell type. Additionally, it provides information about the target mRNA's subcellular distribution, thus offering the possibility to study intracellular transport of particular mRNAs.     

In situ hybridization methods for the detection of somatostatin mRNA in tissue sections using antisense RNA probes

In situ hybridization studies with [32P] and [3H] labelled antisense RNA probes were undertaken to determine optimal methods of tissue fixation, tissue sectioning, and conditions of hybridization, and to compare the relative merits of the two different radioactive labels. The distribution of somatostatin mRNA in neurons of rat brain using a labelled antisense somatostatin RNA probe was employed as a model for these studies. The highest degree of sensitivity for in situ hybridization was obtained using paraformaldehyde fixation and vibratome sectioning. Optimal autoradiographic localization of mRNA was obtained within 7 days using [32P] labelled probes. However, due to the high energy emittance of [32P], precise intracellular localization of hybridization sites was not possible. [3H] labelled RNA probes gave more precise cellular localization but required an average of 18-20 days autoradiographic exposure. The addition of the scintillator, PPO, decreased the exposure time for the localization of [3H] labelled probes to seven days. We also report a method for combined in situ hybridization and immunocytochemistry for the simultaneous localization of somatostatin in mRNA and peptide in individual neurons.     

Nonradioactive in Situ Hybridization Histochemistry in Leukemic and Nonleukemic Culture

Technical limitations are associated with conducting successful in situ hybridization. In this study, three cell types including a tumor neuroblastoma cell line (Neuro-2a), an oligodendrocyte primary culture, and a nonneuronal acute lymphoblastic leukemia cell line (Reh) were used to conduct successful nonradioactive in situ hybridization. Two cDNA probes were used. A 1 kb probe was used to identify the expression of proteolipid protein (PLP) mRNA in a primary culture of oligodendrocytes. A 760 bp cDNA was used to identify the expression of ubiquitin C-terminal hydrolase (UCH-L1) mRNA in Neuro-2a and Reh cells. The probes were labeled with digoxigenin-11-dUTP, denatured, and hybridized with cells fixed on coverslips. The efficiency of the labeling was tested using dot blot analysis by comparing the intensity of our labeled probes with known concentration of the probe labeled by the provider. The nonspecific signals were washed off, followed by detection of a signal specific to the gene. The specificity of the probes was determined by treating the cells with RNase A, hybridizing with bacterial Dig-labeled cDNA (pBR322) and hybridizing the tissues in the absence of labeled probe. During the labeling step, we found that addition of co-precipitants, such as tRNA or glycogen, during precipitation of the labeled probe followed by overnight incubation at -20 C is essential for good recovery of labeled cDNA. Dissolving the labeled probe in a buffer solution containing sodium dodecyl sulfate improves the quantity of the labeling. At the cellular level, prehybridization treatments optimize the permeability of the cell and allow efficient penetration of the labeled probe. Fixing with paraformaldehyde or an ethanol-acetic acid mixture can preserve the structure of cultured cells. To increase the signal to noise ratio, cells were treated with 0.2 N HC1 followed by extensive washes using a solution with a high salt concentration and containing dextran sulfate. This treatment significantly improves the signal and reduces the background in cell cultures, but not in tissue sections. The ability to reuse the labeled probe-hybridization mixture is another advantage for using nonradioactive in situ hybridization.     

In situ hybridization technique to localize rRNA and mRNA in mammalian neurons

In situ hybridization provides a method for identifying cells that contain specific nucleic acid sequences. This report outlines an in situ hybridization procedure for mammalian neural tissue. The method maintains morphological quality and produces excellent specificity. Seven tritiated nucleic acid probes were examined: two ribosomal RNA probes, a control pBR322 plasmid probe, two probes encoding portions of the gene for oxytocin, one probe each encoding a portion of vasopressin glycoprotein, and neurophysin. Using cryostat-cut rat brain sections, rRNA probes labeled the cytoplasm of all cells and the nucleoli of larger neurons. The plasmid probe failed to produce a strong signal. Oxytocin and vasopressin probes appropriately labeled the cytoplasm of hypothalamic magnocellular neurons. Vasopressin parvocellular neurons were not identified by the current method, and the shorter length neurophysin probe failed to produce a signal. Methodological variables were examined by counting autoradiographic grains in cells. The longer oxytocin probe produced a stronger signal than the shorter oxytocin and vasopressin probes, and higher probe concentrations resulted in stronger signal. Hybridization could be abolished by tissue pretreatment with RNAse A, and longer exposure time increased signal strength. The outlined fixation steps with fresh-frozen tissue produced a superior signal compared to paraformaldehyde-perfused tissue.     

In situ hybridization and immunocytochemistry in serial sections of rabbit skeletal muscle to detect myosin expression

We performed in situ hybridization of myosin heavy-chain (MHC) mRNA on rabbit muscle using a biotin-labeled complementary RNA probe. An 1107-nucleotide fragment from an alpha-cardiac MHC cDNA was used to transcribe an RNA probe 97% similar to slow-twitch and 75% similar to fast-twitch sequences. Serial sections were used to identify slow-twitch fibers in medial gastrocnemius, soleus, and tibialis anterior by immunofluorescence of slow MHC and oxidative capacity by histochemistry. Slow-twitch fibers hybridized by the RNA probe stained heavily after detection with streptavidin-alkaline phosphatase (89% dark and 11% medium density). Fast-oxidative fibers stained intermediately (26% dark, 58% medium, and 16% light) and fast-glycolytic fibers stained lightly (12% medium and 88% light). Biotin-labeled probe and enzymatic detection allowed greater resolution of the subcellular location of the MHC mRNA, a distinct advantage over isotope labeling and autoradiography. A non-uniform distribution of MHC mRNA was recognized within an adult skeletal muscle fiber. High concentrations of MHC mRNA were found under the sarcolemma and between the myofibrils, suggesting the existence of a distribution mechanism. The combination of in situ hybridization and immunocytochemistry allows rapid subcellular localization of both MHC mRNA and its translated protein.     

Complementary neuronal and glial expression of two high-affinity glutamate transporter GLT1/EAAT2 forms in rat cerebral cortex

The glutamate transporter GLT1 is essential in limiting transmitter signaling and restricting harmful receptor overstimulation. It has been shown recently that GLT1 exists in two forms, the generic GLT1 and a 3'-end-spliced variant of GLT1 (GLT1v), both with similar transport characteristics. To differentiate clearly the cellular distribution of both GLT1 forms in the cortex, specific cRNA probes for non-radioactive in situ hybridization were generated and applied to adult rat brain sections. The results were complemented by western and northern blot analyses and by immunocytochemical investigations using specific peptide antibodies against both GLT1 forms. The study confirmed that generic GLT1 mRNA was expressed predominantly in astrocytes and, to a small extent, in neurons, whereas GLT1 protein was detected only in cell membranes of astrocytes. On the other hand, GLT1v mRNA and protein were demonstrated predominantly in neurons and in non-astrocytic glial cells irrespective of the cortical areas studied. A cytoplasmic granular staining of neurons and astrocytes predominated in the demonstration of GLT1v protein. It is concluded that the cellular expression of the two GLT1 forms is complementary. The cytoplasmic vesicular distribution of GLT1v may represent an endogenous protective mechanism to limit glutamate-induced excitotoxicity.     

Differential subcellular localization of particular mRNAs in hippocampal neurons in culture

In situ hybridization was used to assess the subcellular distribution of mRNAs encoding several important neuronal proteins in hippocampal neurons in culture. mRNA encoding GAP-43, a protein that is largely excluded from dendrites, was restricted to nerve cell bodies, as were mRNAs encoding neurofilament-68 and beta-tubulin, which are prominent constituents of dendrites and of axons. In contrast, mRNA encoding MAP-2, a protein that is selectively distributed in dendrites and cell bodies, was present in both dendrites and cell bodies. These results demonstrate that different mRNAs are differentially distributed within individual hippocampal neurons. Taken together with previous findings from other laboratories, our results suggest that only a limited set of mRNAs are available for local translation within dendrites.     

A synthetic oligonucleotide probe encoding for atrial natriuretic peptide detects specific mRNA transcripts in rat heart but not brain using in situ hybridization histochemistry

In situ hybridization histochemical techniques were used in an attempt to demonstrate atrial natriuretic peptide (ANP) messenger RNA (mRNA) in the rat brain. A synthetic oligonucleotide derived from previously reported ANF cDNA sequence was used as a probe. Northern blot analysis of total RNA isolated from rat heart demonstrated that the oligonucleotide recognized a single species of RNA (0.9 kb), a size consistent with previous reports. Rat heart sections revealed dense accumulations of ANF mRNA in the cardiac atria and lesser densities in the ventricles. Rat brain sections hybridized with the same oligonucleotide did not label ANF mRNA accumulations in any neuronal cell bodies. A possible explanation for this latter observation is either sparsely distributed expressing neurons or low expression and high turnover of ANF mRNA in brain.     

Ultrastructural visualization of cytoskeletal mRNAs and their associated proteins using double-label in situ hybridization

We have been able to visualize cytoskeletal messenger RNA molecules at high resolution using nonisotopic in situ hybridization followed by whole-mount electron microscopy. Biotinated cDNA probes for actin, tubulin, or vimentin mRNAs were hybridized to Triton-extracted chicken embryo fibroblasts and myoblasts. The cells were then exposed to antibodies against biotin followed by colloidal gold-conjugated antibodies and then critical-point dried. Identification of mRNA was possible using a probe fragmented to small sizes such that hybridization of several probe fragments along the mRNA was detected as a string of colloidal gold particles qualitatively and quantitatively distinguishable from nonspecific background. Extensive analysis showed that when eight gold particles were seen in this iterated array, the signal to noise ratio was greater than 30:1. Furthermore, these gold particles were colinear, often spiral, or circular suggesting detection of a single nucleic acid molecule. Antibodies against actin, vimentin, or tubulin proteins were used after in situ hybridization, allowing simultaneous detection of the protein and its cognate message on the same sample. This revealed that cytoskeletal mRNAs are likely to be extremely close to actin protein (5 nm or less) and unlikely to be within 20 nm of vimentin or tubulin filaments. Actin mRNA was found to be more predominant in lamellipodia of motile cells, confirming previous results. These results indicate that this high resolution in situ hybridization approach is a powerful tool by which to investigate the association of mRNA with the cytoskeleton.     

Quantitation of casein messenger ribonucleic acid sequences using a specific complementary DNA hybridization probe.

Two highly purified rat casein mRNA fractions were used as templates to synthesize complementary DNA (cDNA) hybridization probes using RNA-directed DNA polymerase isolated from avian myeloblastosis virus. Both of the probes selectively hybridized to RNA isolated from lactating mammary tissue, but not to poly(adenylic acid)-containing rat liver RNA. An analysis of the kinetics of hybridization of the cDNA derived from the 15S casein mRNA (cDNA12S) with their individual mRNA templates indicated that greater than 90% hybridization occurred over a R0t range of one and one-half logs with R0t 1/2 values of 0.0023 and 0.0032 mol s l.-1, respectively. Compared with the total RNA isolated from lactating mammary tissue, these values represented a 166- and 245-fold purification, respectively, of these individual mRNA fractions. Using the 15S casein mRNA as a template, two probes of different lengths and specific activities were synthesized. The deoxyribonucleotide and mRNA concentrations and the temperature of incubation were optimized to obtain either a high specific activity cDNA probe, 330 nucleotides long, which represented approximately 25% of the mRNA or a lower specific activity preparation containing some complete cDNA copies, 1300 nucleotides in length. The Tm of the longer cDNA15S-15S mRNA hybrid was 88.5 degrees C, while that of the short cDNA15S-RNA hybrid was 82.5 degrees C. Following this initial characterization, the cDNA15S probe was utilized for three separate determinations: (1) Analysis of the sequence divergence between mouse and rat casein mRNAs. It was observed that the rate of hybridization of heterologous rat cDNA15S-mouse casein mRNA was only 20% that of the homologous rat cDNA15S-rat casein mRNA hybridization. The resulting heterologous hybrid displayed approximately 17% mismatching compared with the homologous hybrid. (2) Determination of the gene dosage for casein mRNA in normal and malignant mammary cells. In this study, an analysis of the kinetics of hybridization of the high specific activity cDNA15S probe with an excess of DNA isolated from lactating mammary tissue, carcinogen-induced mammary tumors, or rat liver indicated that casein mRNA was transcribed from the nonlification or deletion was observed during tumor formation or the process of mammary differentiation. (3) Quantitation of casein mRNA sequences during normal mammary gland development. RNA excess hybridizations were performed using RNA extracted from either pregnant, lactating, or regressed rat mammary tissue. The concentration of casein mRNA molecules/alveolar cell was found to increase 12-fold from 5 days of pregnancy until 8 days of lactation and then declined to approximately 2% of the maximal level of 79 000 molecules/cell by 7 days after weaning. A coordinate increase was observed in casein mRNA sequences detected by cDNA hybridization and mRNA activity measured in a cell-free translation assay.     

Differential regulation of adenine nucleotide translocators by hypertonicity in the brain

To determine the gene(s) induced by hypertonicity in the brain, we performed a differential display analysis using RNA isolated from isotonic and hypertonic rat astrocytes. One cDNA rapidly up-regulated by hypertonicity was isolated, and the DNA sequence revealed that it was identical to adenine nucleotide translocator (ANT)2. ANT2 protein exchanges intramitochondrial ATP for cytoplasmic ADP. Among three ANT isoforms, only ANT2 mRNA was up-regulated markedly from 1 to 4 h after exposure to hypertonicity. Induction of the mRNA did not require de novo protein synthesis. Furthermore, ADP translocase activity in mitochondria of astrocytes was increased significantly by hypertonicity. To see the localization and regulation of ANT2 mRNA in the brain, we performed in situ hybridization of rat brain after intraperitoneal injection of a high concentration of NaCl. Although there were only weak signals in the control, intense hybridization signals were seen in hypertonic rat whole brain. Microscopic examination showed that ANT2 signals were present in the neurons, as well as glial cells. These results suggest that ANT2 may play a role in brain cells to adapt to the hypertonic environment.     

Expression of collagen type transcripts in chick embryonic bone detected by in situ cDNA-mRNA hybridization

The development of the chick embryonic calvarium, an intramembranous bone, is characterized by direct differentiation of cranial ectomesenchymal cells into osteoblasts without the formation of a cartilage anlage. Collagen biosynthesis remains predominantly as type I in the calvaria. However, in severely calcium-deficient chick embryos maintained in shell-less (SL) culture, cartilage-specific type II collagen is synthesized by the calvaria. Immunohistochemistry localized the cells expressing type II collagen to undermineralized regions of the SL bone. In this study, collagen gene expression in bones of normal (N) and calcium-deficient SL chick embryos was examined at Incubation Day 14 by in situ cDNA-mRNA hybridization. A critical step in the procedure, which used biotinylated cDNA probes, was the selection of fixation conditions which maximized RNA retention and maintenance of tissue morphology. Tissues fixed in modified Carnoy's fixative (58% ethanol, 30% choloroform, 10% acetic acid, 2% formaldehyde) for 2-4 hr at -20 degrees C sectioned well and retained their cell morphology and cytoplasmic RNA. Other treatments important for the procedure included demineralization in 0.25 M HCl and removal of matrix by hyaluronidase digestion. In situ hybridization with type-specific collagen cDNA probes revealed that type II collagen mRNA was present in cells throughout the SL calvaria. More importantly, cells with type II collagen mRNA were also present in N calvaria which do not synthesize the protein. The overall abundance of type II-positive cells in N calvaria was not significantly different from that in SL calvaria, but their distribution throughout the bones differed. In general, the regional distribution of type II cells was inversely correlated with the extent of matrix mineralization. In the N calvaria, cells containing collagen type II mRNA were absent in the extensively mineralized superior zone, but were found in the temporal zone which showed limited mineralization. On the other hand, in the SL calvaria, which were substantially undermineralized overall, cells with type II mRNA were found throughout the tissue. Interestingly, the overall ratio of type I cells to type II cells was approximately 50% higher in N calvaria. These findings suggest that collagen type mRNA expression in the chick embryonic calvarium is correlated with, and perhaps dependent on, the extent of tissue matrix mineralization.     

Localization of actin messenger RNA during early ascidian development

The spatial distribution of RNA sequences during early development of the ascidian, Styela plicata, was determined by in situ hybridization with poly(U) and cloned DNA probes. Styela eggs and embryos contain three colored cytoplasmic regions of specific morphogenetic fates, the ectoplasm, endoplasm, and myoplasm. These cytoplasmic regions participate in ooplasmic segregation after fertilization and are distributed to different cell lineages during early embryogenesis. n situ hybridization with poly(U) suggests that poly(A)+RNA is unevenly distributed in eggs and embryos, with about 45% in the ectoplasm, 50% in the endoplasm, and only 5% in the myoplasm. In situ hybridization with a histone DNA probe showed that histone RNA sequences were not localized in eggs or embryos and distributed between the three cytoplasmic regions according to their volumes. In situ hybridization with an actin DNA probe showed actin RNA was localized in the myoplasm and ectoplasm of eggs and embryos with about 45% present in the myoplasm, 40% in the ectoplasm, and only 15% in the endoplasm. These results suggest that a large proportion of the egg actin mRNA is localized in the myoplasm, participates in ooplasmic segregation after fertilization, and is differentially distributed to the mesodermal cell lineages during embryogenesis. Analysis of the translation products of egg mRNA suggests that the localized mRNA codes for a cytoplasmic actin isoform.