Molecular Analysis of the Single Cell: Importance in the Study of Psychiatric Disorders

James Eberwine, Peter Crino, Steve Arnold, John Trojanowski, Scott Hemby

Since Psychopharmacology: The Fourth Generation of Progress was published five years ago, the application of the tools of molecular biology to the study of the central nervous system (CNS) has revolutionized the way in which we think about neuronal plasticity, development, and physiological processes, as well as neurologic and psychiatric illnesses. Such progress has highlighted a number of unresolved issues. One of the major problems that must be addressed during the coming decade is how to study and interpret the biochemistry and molecular biology of individual cellular components (neurons and glia) of the CNS in the appropriate cellular context. The analysis of gene expression in the CNS is difficult, as a result of its vast complexity, which derives from the large number of cell types, the myriad synaptic connections which can be made between neurons (resulting in unique identities for cells which appear phenotypically similar), and the large percentage of the genome (70% of the genes) that is expressed in a cell-selective manner.

Indeed, most techniques of molecular analysis utilize large amounts of cellular mRNA isolated from a tissue block for standard analysis, such as Northern blotting, cDNA library construction, primer extension, etc. This level of resolution is insufficient to determine specific cellular levels of mRNA in selected cells within the tissue block. The basic problem is that the mRNA from the cells of interest is diluted by the mRNA from the surrounding cells, which decreases the specific signal. The most useful technology for addressing this problem to date has been in situ hybridization, This technique has single-cell resolution but can only be used to assess the presence or absence of a few species of mRNA at a time.

To circumvent these problems, we have developed several methodologies which increase the sensitivity of mRNA and protein detection, while requiring only the amount of material obtainable from a single cell. The details of most of these techniques have been published elsewhere so an in-depth discussion is unnecessary (7, 17). However, a brief overview of these techniques and some examples of their application to questions of neurobiological and psychiatric interest will illustrate our view of the direction in which research into the study of the molecular biology of the CNS should proceed.

The RNA amplification procedure (aRNA) was developed to amplify or replicate the mRNA population found in tissue samples and single cells (Eberwine et al., 1993). This is a linear amplification procedure in which a bacteriophage promoter is attached to the double-stranded cDNA copy of each of the mRNAs in a cell. This promoter functions as a recognition site for the binding of a specific RNA polymerase which in the presence of ribonucleotides will synthesize RNA from the double-stranded cDNA template. This procedure linearly amplifies the material from a single cell by a factor of nearly 103 with one round of amplification and 106 with two rounds. Each amplification step is linear, and yields large quantities of amplified products with limited misrepresentation of amplified RNA abundances.

The amplified RNA population can be analyzed using a number of procedures, including the polymerase chain reaction for specific mRNAs, differential display to amplify fractions of the RNA population of cDNA, or cDNA library construction. However, when analyzing a large number of gene products from the aRNA, we developed a procedure called expression profiling to determine the relative levels of expression of multiple mRNAs simultaneously (from one mRNA to several thousand). This procedure utilizes arrayed cDNAs, oligonucleotides, or genes as hybridization targets for the aRNA. When this technique is performed appropriately, the relative levels of hybridization reflect the relative levels of the corresponding RNA in the aRNA population, which in turn reflects the mRNA abundance in the original single cell.

These techniques allow us to ask questions of basic and clinical importance at the single-cell level. Such questions might include 1) What are the changes in gene expression that occur as a cell progresses to an oncogenic phenotype? 2) What coordinate changes in gene expression occur in identified single cells possessing a known receptor repertoire after pharmacological challenge? 3) What are the changes in gene expression in particular neuron types with development, maturation, and aging? In this chapter we will discuss our work in defining differences in mRNA abundances in distinct subcellular sites of neurons (neuronal polarity), a factor that is important in controlling neuronal development. We will also cover our initial efforts to apply these ideas and techniques to the study of schizophrenia.

The complexity of the CNS extends beyond that already described. Neuronal morphology shows an extensive cellular polarity, with one axon sending signals from the cell soma to the next neuron and an extensive array of dendrites receiving input from presynaptic neurons. In particular, neurons contain a mosaic of cytoplasmic and membrane proteins which are differentially distributed in axons, dendrites, and somata. Substantial ultrastructural evidence documents the presence of functional protein synthetic machinery in neuronal processes, especially at or near post-synaptic sites. In individual live, cultured rat hippocampal neurons, the mRNAs for members of various second messenger systems as well as neurotransmitter receptors are present in individual processes of hippocampal cells in culture (15). Most of these studies have been performed on mature neuronal processes, with speculation that local synthesis of the corresponding proteins may play a role in synaptic plasticity.

Recently, we embarked upon studies to examine the complexity of mRNA populations present in the developing dendrite and dendritic growth cone in an effort to determine whether local protein synthesis of specific mRNAs may play a role in synaptogenesis (6). For these studies, we utilized the dispersed hippocampal primary cell system developed by Dr. Marc Dichter (4). Prior to amplification of the material from individual dendrites, we stained a set of these cells with acridine orange to assess whether these processes indeed contain mRNA. Under the appropriate conditions, acridine orange will stain single-stranded RNA an orange color, while double-stranded DNA will be stained green. The data presented in the photomicrograph in Figure 1shows a light orange staining in the apical dendrite of the centered neuron. Data such as these establish the presence of RNA in the dendrites of the cultured cells used in our studies.

The hippocampal cells were next assayed for the presence of specific mRNA subtypes using aRNA amplification and expression profiling. These data have shown a much expanded repertoire of dendritically localized mRNAs (Figure 2), compared with what was previously known. Among the mRNAs present in growth cones are those encoding various intermediate filaments, as well as various growth factor receptors. The relative levels of mRNAs can vary between different growth cones as well as between the dendrite and cell soma. For example, the relative levels of trkA to p75 (low affinity NGFR) in the dendrite is approximately 3:1 (Figure 2), whereas in the cell soma the ratios is different 0.8:1 (data not shown). These data illustrate an important concept, namely that the relative levels and coordinate changes in the expression of multiple genes underlie normal cellular homeostasis as well as a cell's response to a stimulus.

The characterization of multiple mRNAs in neuronal processes suggests that subcellular localization of mRNA, mRNA transport, and local protein synthesis may regulate aspects of neuronal physiology. Indeed, in recent studies we have shown that protein synthesis can occur in individual neuronal dendrites using transfection of single dendrites with lipid-encoated mRNAs (6). Protein synthesis from the transfected mRNA was detected using antibodies. These data further emphasize the potential importance of mRNA localization in normal neuronal functioning.

Such localization issues may also be important for understanding abnormal neuronal functioning in psychiatric illnesses. For example, an imbalance in dopamine responsiveness is thought to underlie aspects of schizophrenia. If dopaminergic activation or inhibition elicits an alteration in mRNA/protein transport or protein synthesis in dendrites, then studies directed towards regulating this particular site of drug action may be more therapeutically effective then targeting of pharmaceuticals to other cellular sites.

Schizophrenia is a multigenic disease whose biological underpinnings have remained elusive (11, 14, 16). Numerous studies have examined various aspects of the disease, with hopes of finding pharmacological agents which will control schizophrenic symptomatology. Abnormalities in a variety of brain regions have been postulated in schizophrenia, with the limbic cortex (entorhinal cortex, cingulate gyrus, and hippocampus), limbic striatum, and prefrontal neocortices being most commonly highlighted (3, 10, 11).Fig. 3 Antipsychotics, primarily dopamine receptor antagonists, are the best long-term treatment so far, but they leave many disabling symptoms of schizophrenia untouched (13). Dopamine receptor blockers such as clozapine (an atypical neuroleptic) and haloperidol (a typical neuroleptic) ameliorate some symptoms but their mechanism(s) of action are unclear. The efficacy of antipsychotics suggests that dopamine neurotransmitter function is altered in schizophrenics, leading to the formation of the dopamine hypothesis of schizophrenia. In an effort to understand the functional role of dopamine in modulating schizophrenia, investigators have turned towards systems which are more amenable to experimental manipulation than those of the human: these include studies in cell cultures and in the rat and mouse. Topics of these studies have included dopamine receptor binding, control of second messenger systems, dopamine-induced changes in the electrophysiologic status of cells, and determination of the role of dopamine in regulating gene expression in those CNS regions thought to be altered in schizophrenia. Various mRNA levels have been shown to be altered in response to regulation of dopaminergic pathways, including proenkephalin (1, 19), c-fos, and GAD67 (2). However, these studies have not provided a cogent hypothesis for the mechanism of action of dopamine receptor agonists/antagonists in normal or abnormal (schizophrenic) neuronal functioning.

In part, this has been difficult because of the multigenic nature of schizophrenia. Indeed, as discussed above, it is likely that the coordinate expression of multiple genes is required for normal as well as abnormal cellular function (7). Many mRNAs are expressed in a cell-specific manner, yet when mRNA is isolated from a tissue, the mRNA present in specific cell types is diluted by the mRNA contribution of the surrounding, non-expressing cells. This may obscure cell specific alterations present in schizophrenia. To address the issues of cell specificity of expression and coordinate changes in gene expression in schizophrenia, we have used aRNA amplification and PCR differential display (12) to determine the mRNA composition of single, immunohistochemically stained entorhinal cortex neurons from schizophrenics and age-matched controls. These sensitive methodologies have coupled the molecular biology of individual human neurons with their neuroanatomy and physiology.

Initially, immunohistochemistry was performed on tissue sections from entorhinal cortex taken from schizophrenics and age-matched controls. The technique utilized an antibody which recognizes the NFM antigen present in pyramidal and stellate neurons. Next, in situ transcription using oligo-dT T7 primer was performed on these sections so that the mRNA present within these sections was converted into cDNA. Using a microcapillary, cDNA was harvested from individual NFM-positive neurons after visualization of the immunostained cell under the microscope (5). This cDNA was amplified into aRNA, the aRNA was converted to single-stranded cDNA, and differential display was performed on this cDNA. As can be seen in Figure 4, we have generated differential display patterns from individual cells from two different schizophrenics (Sch 1 & 2) and two age-matched controls (AM 1 & 2). A control for the PCR reaction was also done in which no template was added to the PCR reaction, so any bands appearing in these lanes (NT) result from amplification of primer sequences.

The data in Figure 4 serve to highlight several aspects of this type of molecular analysis. Many of the differential display (DD) bands are common to all four individuals, and these bands likely correspond to mRNAs which are either common to this cell type or are common to all cells (such as those which encode proteins involved in basal cellular function). There are also bands which appear to be enriched in the samples from schizophrenics, while other bands appear to be enriched in the age-matched controls. The difference in abundances of these bands between the samples suggests that the bands which are enriched in the cells from schizophrenics may correspond to mRNAs which are up-regulated in response to schizophrenia, while the bands enriched in the normal samples would correspond to mRNAs which are down-regulated in schizophrenics. There are, of course, complications to this interpretation—including biological uncertainties such as the drug history of the donors, inherent differences in morphologically similar cells or even appropriate choice of cells for analysis. Technical problems, including the non-linear amplification produced by PCR, must also be considered. With these caveats in mind, it is still reasonable to take the differentially displayed bands and treat them as potential markers for schizophrenia. It will be necessary to characterize them biologically in order to determine their identity, cell specificity of expression, and whether the corresponding mRNAs are indeed differentially represented in schizophrenics versus age-matched controls. It would also be prudent to use them as probes to determine whether they map to any of the genetic loci currently being mapped and characterized for families of schizophrenics. As a result of such studies, both expression profiling and differential display, it should be possible to characterize genes associated with the schizophrenic phenotype.

It would be reasonable to speculate that coordinate changes in gene expression resulting from schizophrenia might provide insight into the underlying cause(s) of schizophrenia. The mRNA changes occur because of changes in transcription rate and mRNA stability; the resulting relative levels are a reflection of these cellular operations. If these relative mRNA levels can be adjusted so that the levels present in disease are returned to those found in nonschizophrenic individuals, such changes may be therapeutic. This idea is called "Transcript Aided Drug Design" and may be useful in formulating pharmacological manipulations to convert disease phenotype into nondisease phenotype. This discussion of the molecular biology of schizophrenia relies primarily upon analysis of the mRNA complement of the neuronal soma. The earlier discussion of mRNA localization makes it intriguing to speculate that changes in mRNA localization (as reflected in dendritic expression profiles) or translation of these mRNAs locally in dendrites may cause aberrant dendritic functioning. Dysregulation of these steps in neuronal functioning has not yet been examined as a possible cause or consequence of schizophrenia.

The studies we have outlined show the power as well as the promise of the application of single cell molecular biological analysis to the study of neuropsychiatric disorders. It is safe to speculate that within five years a detailed molecular fingerprint of schizophrenia will be available which could prove invaluable as a means of early detection and, ultimately, as a tool to genetically alter the course of this psychiatric illness. The challenge in the coming years will be to determine how to analyze the information provided by this molecular fingerprint to develop therapeutics in an effort to benefit the patient.

This work was supported by NIH AG9900 and a NARSAD Established Investigator Award to JE, PC is a Howard Hughes Postdoctoral Fellow and SH is supported in part by a NRSA from NIDA and a NARSAD Young Investigator Award.

published 2000