Background

Early research was in bacterial genetics and the control of transcription in E. coli, with emphasis on histone-like proteins and the firA gene (commentary: Dicker and Seetharam, 1992).  After studying gene expression in Drosophila, efforts focussed on recombinant DNA and oligonucleotide technology (eg. Optimised probe screening). A new programme of research was the exploitation of recombinant vaccinia virus for the development of new vaccines, first against rabies and subsequently to papillomavirus-induced tumours.  The recombinant rabies vaccine (Kieny et al., 1984; patents) has been extensively used in Europe and North America for the eradication of rabies in the wild (V-RG/Raboral vaccine).  The method was applied to anti-tumour vaccination (Lathe et al., 1987; patent) focussing on papillomavirus antigens E5, E6 and E7 (Meneguzzi et al., 1991), breast cancer antigens (pS2, hSP and epithelial tumor antigen MUC-1: Transgene, 2002). Recombinant expression was extended to transgenic animal technology, with new applications in pharmaceutical production (Lathe et al., 1986; patent) and the analysis of brain function (Workshop 1992; Lathe and Morris, 1994).

Hippocampal function

The function of the mammalian hippocampus is a topic of much debate.  It plays a central role in memory (Scoville and Milner, 1957; Squire, 1995) and is an early site of dysfunction in Alzheimer’s disease.  However, its exact role is unknown, and others have attributed a much wider role to the formation (see Kimble, 1968; Jarrard, 1995).  We looked at genes selectively expressed in the hippocampal formation, not only with the hope that these might cast light on the function it plays, but also to provide tools for region-specific transgenesis (e.g. Lathe, 1996); three techniques were successful (reviewed by Pickard et al., 1999).

Differential hybridisation: CYP7B and steroid metabolism

This highlighted a new cytochrome P450, CYP7B, richly expressed in hippocampus (Stapleton et al., 1995), see Figure 1.  In association with Jonathan Seckl (Edinburgh) and David Russell (Dallas) we demonstrated that the cloned enzyme hydroxylates brain steroids at the 7alpha position (Rose et al., 1997), in addition to modifying cholesterols in liver (Rose et al., 1997; Schwarz et al., 1997; Martin et al., 1997). 

CYP7B substrates include estradiol, important for its neuroprotective actions, and the major adrenal steroid in primates, dehydroepiandrosterone (DHEA), whose asymptotic decline with age has suggested a causal role in human cognitive ageing and AD.  7-oxygenation of DHEA is likely to represent activation of the molecule. This is confirmed by reports from Lardy's lab.

We prepared mice lacking CYP7B enzyme (collaboration with Jonathan Seckl, Edinburgh and Jan-Ake Gustafsson, Stockholm) with a chromogenic tag under control of 7B gene transcription (Rose et al., 2001).  Intense chromogenic staining in hippocampus is accompanied by abolition of 7-hydroxylation and a small deficit in spatial learning.

We recently reported expression of CYP7B in human hippocampus and downregulation in Alzheimer’s disease (Yau et al., 2003). Elsewhere, CYP7B modification of androstanediol gates access to target receptors including ER and AR (Weihua et al., 2002); a similar mechanism may operate in hippocampus.

Brain serine protease 1 (BSP1)

We explored 8 gene families for members specific to the hippocampus.  Serine proteases were selected because of the known association between proteolytic cleavage of the amyloid precursor protein (APP) and Alzheimer’s disease.  Family-specific PCR  amplification showed  that  the  predominant  serine  proteases in rodent hippocampus are tissue-type plasminogen activator (t-PA); RNK-Met-1, a lymphocyte protease not reported in brain; and two new family members, BSP1 (brain serine protease 1) and BSP2 (Davies et al., 1998).  The t-PA substrate plasminogen was wholly absent.  The new enzyme BSP1 (also known as neuropsin and KLK8) is selectively expressed in hippocampus (Davies et al., 1998).

To explore the role of BSP1 we prepared mutant mice lacking the enzyme.  These animals showed no changes in hippocampal-dependent behaviours, but there was clear evidence for polyspiking activity (collaboration with Ceri Davies, Edinburgh now Glaxo) and a striking hypersensitivity to the glutamate agonist, kainic acid (B. Davies et al., 2001).  Administration was not tolerated by the mutant animals, and brain sections showed a significant increase in cortical activation.

This result confirms a role for the hippocampus in cortical activation, and further argues that BSP1 normally acts to suppress activity in other brain regions.  Intriguingly, a very recent report from a Canadian group (Shimizu-Okabe et al. 2001) indicates that BSP1 is at least 10-fold upregulated in Alzheimer's: perhaps overexpression of BSP1 is linked to failure of cortical activation. 

Gene-trapping

In a third approach a promoter-less reporter was inserted into random genes within ES cells and ensuing mouse lines examined for expression in adult brain (Steel et al., 1998).  Unexpectedly, a large proportion of gene-trap lines expressed reporter activity in hippocampus. 

Three insertions were examined in detail.  All identified membrane-associated signalling molecules.  One interrupts a macrophage-inhibitory protein (MIP1) receptor known as EBI-1, and mice homozygous for this disruption displayed abnormal long-term potentiation (LTP) in the hippocampus (Steel et al., 1998; project in collaboration with W. Skarnes [Edinburgh now Berkeley] and R. Morris [Edinburgh]).  The question is prompted - why should an immune mediator such as MIP1 affect hippocampal LTP?

Implications for function

Statistical analysis of the gene-trapping study argues that [with 95% confidence] ~1/3 of mouse genes are expressed in the hippocampus.  The gene-trap products analysed were cell surface signal transduction molecules (eg. MIP1 receptor), meshing with the identification of CYP7B (steroid hydroxylase).  A general survey (Lathe, 2001) suggests that the hippocampus (like the hypothalamus) contains one of the highest densities of receptors for soluble ligands in the brain: these are likely to be accessible, functional, and mediate physiological and cognitive changes in vivo.  Thus it seems that the hippocampus is a primary target for ligands that reflect body physiology including ion balance and blood pressure, immunity, pain, reproductive status, satiety and stress.  A primary role of the early hippocampus, an offshoot of the olfactory system, could have been in sensing hormones and metabolites in blood and cerebro-spinal fluid. 

This does not address the final output of the hippocampus.  O.S. Vinogradova (1975; recently revisited in Hippocampus, 2001) proposed that the hippocampus computes novelty by comparing sensory and memory inputs.  Enteroceptive modulation could permit the hippocampus to compute salience rather than just novelty.  Further, the formation can act as an endocrine transducer, and govern adaptive changes in body physiology and hypothalamic/pituitary/adrenal (HPA) axis activity.  Glucocorticoid levels rise in response to a novel stimulus, or to a previously learned aversive taste, the hormonal rise is abolished by hippocampectomy.  This meshes with the inferred role of the hippocampus in anxiety (Gray, 1982; McNaughton and Gray, 2000)

Adrenal hormones (glucocorticoids and norepinephrine) enhance consolidation of memory traces, endocrine activity directed by the hippocampus may explain its role in memory.  In support, the requirement for the hippocampus in memory acquisition can be circumvented, at least in part, by co-administration of adrenal hormones (reviewed by Lathe, 2001).  This view differs from that proposed traditionally (and pertaining to the role of the hippocampus in spatial navigation) but the two are mutually consistent. 

In collaboration with Neil McNaughton (Otago, NZ), we have been developing a further theory in which the hippocampus, under hormonal modulation, (a) directs selective cortical activation (as demonstrated by the properties of our BSP1 knockout mice, Davies et al., 2001) and (b) gates voluntary behaviour (to be published).

Hydroxylated steroids in yeast

A study has recently been completed in collaboration with Eric Degryse (Strasbourg),in recombinant yeast was engineered to produce 7-hydroxylated DHEA.  This involved several genomic modifications, including the inactivation of a yeast acetyl transferase (ATF2) and a potent 17-hydroxysteroid hydroxylase activity (AYR1), as well as the overexpression of rat CYP7B enzyme (Vico et al, 2002).

Targets for 7-oxygenated steroids remain obscure.  We hypothesise that 7-hydroxyDHEA and related molecules may exert biological effects by targeting intracellular receptors for sterols, including the emopamil binding protein (EBP), the sigma site, and the peripheral benzodiazepine receptor (PBR).  These molecules are involved in the synthesis and metabolism of cholesterols and ligands modulate cell life and death in the CNS; they also bind steroids (discussed by Lathe and Seckl, 2002; Lathe, 2002).  More recently, an important effect on microtubule function has become apparent (Murakami et al., 2001).

Eurosterone Chip

The Edinburgh Eurosterone consortium (Steve Hillier, R. Lathe, John Mullins, Ian Mason, Jonathan Seckl, with Alex Bailey, Adminstrator), together with several European partners, is assembling a DNA array to comprise the majority of known steroid-metabolising and target cDNAs.  The completed chip, that recently became available, comprises ~1000 genes from mouse, rat and human, gridded out (70-mer oligonucleotides) for hybridisation (to be published).  First results have shown dramatic changes during ovarian stimulation (in press).

Gene expression variegation

Additive transgenes are subject to sporadic fluctuations in expression level that depends on copy number (Mehtali et al., 1990), a complicating factor for studies employing transgenic animals. We found, in mammary gland, that expression level appeared to be clonal areas: patches of cells with high expression flanking other zones where expression was absent - the extent of variegation also depended on the integration site (Dobie et al., 1996; 1997). Further, in a transgene homozygote, only one allele was expressed (mono-allelic transgene expression; Opsahl et al., 2003).

DNA hybridization parameters

We reassessed melting temperature and find that the standard Schildkraut-Lifson equation is inaccurate for immobilised nucleic acids (Rose et al., 2002).

Regulated gene expression in the hippocampus

Long-term gene expression changes in the CNS underlie physiological changes including memory, and pathologic states such as depression.  Most studies have examined stimulus-induced gene expression over a period of hours. To address changes over a longer time period (days), in a collaboration with Ian Reid (Aberdeen) we examined rats subjected to electroconvulsive stimulus (ECS; Reid and Stewart, 1997).  mRNA prepared 72 hours (3 days) post-treatment was used to prepare cDNA.  Subtractive hybridization was performed between ECS and sham-treated cDNA pools to identify genes differentially expressed following ECS. 

Though many clones were strikingly and reproducibly enriched (>100 fold) in the ECS cDNA population, corresponding mRNA levels were only marginally increased (<5 fold).  This suggests a gene-specific ECS-sensitive barrier to cDNA synthesis.

ECS-upregulated cDNAs included beta-catenin, phosducin-like protein (PhLP), tomosyn, and ubiquitin-specific protease Usp7.  These genes are of enormous interest - beta catenin is a major target for mood-stabilising drugs, and the major changes in beta-catenin mRNA availability and abundance suggests that ECS (ECT in human) and pharmacologic agents act through common pathways.

Nature of the Scrapie Agent

The causal agent for the scrapie-like diseases including CJD in human is not fully understood.  It has been suggested that a major component of the agent may consist of a proteinaceous replicating derivative of the cellular polypeptide, PrP (Prusiner, 1982).  However, the protein-only hypothesis fails to explain the existence of distinct strains of disease, with defined characteristics, that propagate through different hosts while retaining the distinctive features of the parent strain.  It has been suggested that PrP, a nucleic-acid binding protein, may complex with a nucleic acid component that provides strain-type characteristics (Weissmann, 1991).  This hypothesis has been emphasised by the recent finding that host nucleic acid modulates the in vitro replicative properties of PrP (Deleault et al., 2003).

In a collaboration with Peter Estibeiro and Eleanor Barnard (ExpressOn Biosystems) we have analysed RNA species differentially expressed in scrapie-infected and sham hamster brain (samples courtesy of the IAH Neuropathogenesis Unit, Edinburgh).  We find that some cellular RNAs are selectively over-represented in scrapie brain (to be published).

Origin of life

Consideration of the replicative properties of small nucleic acids prompted re-evaluation of current notions of the origin of terrestrial life.  Replicating prebiotic polymers are thought to predate the emergence of true life-forms, but the initial mode of replication is unknown.  However, it demands an explanation based on local physicochemistry.  Dual consideration of the conditions of the early terrestrial surface, with the unusual physicochemical properties of nucleic acids like DNA, could explain the emergence of nucleic acids as key biomolecules.  The early impact that produced the moon, and fast terrestrial rotation, subjected coastal areas 3.9 Ga ago to rapid tidal flooding (dilution) and drying (concentration), with a likely periodicity in the range of 2-6 h, and could have provided a driving force for cyclic replication of early biomolecules.  Such a mechanism applies only to molecules capable of association /polymeri\ation at high salt concentration, and of dissociation at low salinity.  Nucleic acids meet these criteria.  It is suggested that tidal cycling, resembling the polymerase chain reaction (PCR) mechanism, could only replicate and amplify DNA-like polymers.  This mechanism suggests constraints on the evolution of extra-terrestrial life (Lathe, 2004).

Autism and the limbic brain

Autism is an early onset disorder characterized by language delay, socio-emotional deficits, and repetitive behavior, often accompanied by anxiety and epilepsy.

Structural and functional studies implicate the hippocampus and amygdala (H+A).  The deficits of autism are consistent with the known role of H+A in behavioral organization, social interaction, epilepsy and anxiety.  Less well known are the neuroendocrine abnormalities of autism, with disturbances of the brain-adrenal and brain-gonadal axes, accompanied by immunological and gastrointestinal dysfunction.  Subtle changes are seen in the regulation of oxytocin, cortisol, androgens and interleukins. 

Together these suggest perturbations in the complex feedback circuitry linking brain and behavior with hormonal and physiological changes that, in turn, feed back to the H+A ('cognitive physiology').  Central to this circuit is the developmental substructuring of H+A that may depend on continuing neurogenesis (prominent in the dentate gyrus) and imprinting maintenance, suggested by autistic features of Rett syndrome.  One imaging study has prominently highlighted dentate abnormalities in autism (Saitoh et al. 2001), and children with bilateral hippocampal lesions meet the diagnosis of autism (DeLong and Heinz, 1997).

There is debate concerning the prevalence of autism.  A 273% increase has been noted between 1987-1998 in California.  Some recent estimates in the US and the UK point to 1/160 children though increasing diagnostic acceptance may contribute (MRC, 2001).  The number of children now presenting with autism is of enormous concern, and the rise could point to a environmental determinant.

Suspicions concerning the role of heavy metals were first raised by Hallaway and Strauts in their book 'Turning Lead into Gold' who reported  elevated lead and other heavy metals in affected children, and the benefits of chelation therapy.  Increased blood lead in autism was first reported in the 1970s, but was ascribed to unusual eating habits of autistic children who often nibble on non-food items, and risk ingesting lead paint for instance.  But this odd nibbling behavior, termed 'pica', is a well-known feature of trace metal deficiency that accompanies metal poisoning. 

Mercury is also implicated: Bernard et al. (2001) emphasised the superficial similarities between mercury poisoning and autism - repetitive behaviours, mental impairment, predisposition to epileptic seizures.  Recently, Holmes et al. (2003) suggested that autistic children may have a deficit in export / detoxification of heavy metals, particularly mercury.  Though this primary result remains to be confirmed, positive results have been obtained with chelation therapy (Holmes, 2003).

A role for heavy metals is plausible, given the likely involvement of the limbic brain in the pathoetiology of the disorder:  the hippocampus contains one of the highest densities of ‘natural’ heavy metals in the CNS.

Further research will be directed to an understanding of the genetic predisposing factors in autism and the mechanisms by which heavy metals might lead to limbic dysfunction.