1.5 Molecular aspects of hormonal regulation

The wide molecular effects of hormones have complicated the understanding of how hormones work on a cell. The old view was of a linear signalling pathway from the receptor to the nucleus, thereby stimulating gene transcription. This view is probably an oversimplification. Hormones can not only regulate most of the molecular machines of the cell, certainly the transcription machinery, but also others. These machines perform and coordinate functions such as RNA and protein biosynthesis, macromolecular transport, cell division or death, and intracellular signalling. Physiological studies have shown that hormonal regulation is specific, yet flexible, and has the ability to generate feedback loops. Advances in genetics, cellular, and molecular biology, and biochemistry have allowed much new, and sometimes confusing, data on the mechanisms underlying hormonal regulation. Many advances have been due to methods of identifying and verifying networks of interactions between proteins. One example is the yeast two-hybrid system, an in vivo genetic screening method for such interactions. Another example is the use of protein tagging (e.g. with histidine residues) which can allow rapid and high-yield protein purification for biochemical studies. This chapter will briefly review some of the mechanisms of hormonal regulation.

The control of the expression of a pattern or network of genes is an essential mechanism for the maintenance of stable, but specific, cellular state or differentiation. The hormonal milieu is often vital to the upkeep of specialized cellular functions, in both cultured cells and in tissues. Many hormones achieve these effects in a cell mainly by enhancing or silencing the transcription of specific genes (see Chapter 1.4 for details about transcription). The molecular details of several eukaryotic transcriptional enhancer systems have been elucidated (for a review, see Ogata et al. (1)).

The general principles are that transcriptional activation involves the formation of a multiprotein complex called an enhanceosome, which assembles at enhancer DNA sequence elements (Fig. 1.5.1). The enhanceosome complex attracts and engages the basal transcriptional machinery on several of its protein surfaces. The surface of the complex is made up of several peptide domains, either from different proteins, or from the multiple domains of one protein. The structure, shape, and domain components of this surface partly explain some of the specificity in transcriptional regulation. Only regulator protein domains that fit together can function together. Another principle is that the attraction between proteins within and between complexes is reciprocal. Thus, binding is cooperative: proteins that fit together help each other to bind to the activator complex. Transcription is stimulated in a synergistic (not additive) manner, once cooperative and specific binding has allowed a threshold of activator concentrations to be reached.

Proteins in the enhanceosome have several functions. Some are DNA sequence-specific transcriptional activators, which bind chromatin and hyperacetylate histones. This chemical modification results in the disruption of chromatin and histone structure around the DNA. The change in chromatin structure is required for transcription to occur. Otherwise, the transcription complex fails to build properly and cannot progress along the DNA template to make pre-mRNA. Other proteins are architectural proteins, which bend the DNA to allow stabilization of the complex and further recruitment of activators. One example is the high-mobility group I(Y) protein (HMG I(Y)), which bends DNA to facilitate the binding of the transcriptional activator, nuclear factor kappa B (NF-κB). The interferon (IFN) β enhanceosome contains several other proteins, including members of the interferon regulatory factor family (IRF1, IRF3, IRF7) and activation transcription factor 2 (ATF2), and is capable of interactions with the basal transcriptional apparatus. Kinase-signalling pathways can modify the activator proteins in the enhanceosome and this phosphorylation stimulates cooperative binding and complex assembly. The enhanceosome now interacts with the basal RNA polymerase II transcriptional apparatus at the gene promoter. The multiple contacts between the surfaces of the enhanceosome complex and the basal transcription complex allow reciprocal strengthening of the stability of a higher order transcriptional complex. This is associated with the recruitment of more coactivator proteins (p300/CBP). The recruitment of p300/CBP (probably by interactions with IRF3) is associated with dramatic hyperacetylation of histones H3 and H4 at the site of the IFNβ promoter. Thus, specificity of transcription (i.e. localization of function at the IFNβ promoter) is achieved by the multiple stepwise interactions required to assemble the higher order transcriptional complex. Once the transcription complex is assembled, and the chromatin structure opened, transcription is activated in an exponential and synergistic manner in response to the external stimulus.

Hormones, such as glucocorticoids, may repress, as well as activate, gene expression. Some of the mechanisms of action of the glucocorticoid receptor (GR) have been elucidated by genetic technology in the mouse. In particular, these experiments have allowed the dissection of GR-mediated gene activation and repression (3). Mice have been made where the gene for the GR has been knocked out (GR null). The mice die after birth because of pulmonary atelectasis, but also show marked adrenal hyperplasia, high corticosterone levels, and reduced expression of gluconeogenic enzymes in the liver. Mice have also been engineered with a single mutation in the D-loop, which forms part of the DNA-binding domain of the GR. This mutation (GR dim) destabilizes the dimerization of the GR, thereby reducing cooperative binding between two GR monomers required for interactions with DNA. The GR dim mice show loss of functions that require dimerization and DNA binding. Conversely, functions that are mediated by interactions between GR monomers and other proteins are preserved. Comparison of the phenotypes of the two mutant mice, GR null and GR dim, and wild-type mice has allowed a functional classification of GR mechanisms (Table 1.5.1, based on Karin (4)). Examples of the positive transcriptional functions, requiring DNA binding by the GR, include transcription of liver genes such as tyrosine aminotransferase, and the viral MMTV genome. Transcriptional repression that requires DNA binding occurs in the proopiomelanocortin and prolactin genes. Finally, a set of functions not requiring DNA binding (and presumably due to interference with other transcription factors by protein–protein interactions) includes repression of the transcription of the osteogenic collagenase and gelatinase genes. The viability of the GR dim mice, in contrast to the lethality of the GR null mutant, reflects the importance of the transcriptional interference functions of the GR, particularly in adrenal and lung development. Additionally, these experiments opened avenues whereby treatments can be targeted to specific aspects of glucocorticoid function. Hopefully, drugs in development will retain the immunosuppressive effects of glucocorticoids (by interfering with the NF-κB signalling system), but without repressing transcription of osteogenic genes and thus avoiding glucocorticoid-associated osteopenia (5).

After transcription, the newly made pre-mRNA must undergo several processing steps in order to be exported from the nucleus and made into protein in the cytoplasm. In humans and higher eukaryotes, substantial variations in the mRNA are introduced during these steps, such that one gene often encodes for several mRNA isoforms. Thus, the final pattern of cellular gene expression is very different from the genomic DNA sequence (6). Hormones can influence pre-mRNA processing mechanisms (7). Many examples have been reported where changes in the ratios of mRNA products from the same pre-mRNA gene transcript occur in response to changes in hormonal conditions (Table 1.5.2). Several processing steps may be involved, including pre-mRNA splicing (8), polyadenylation, and turnover (9). In some cases, the sequence elements on the pre-mRNA through which regulation is mediated and proteins acting as regulators have been identified (10).

Alternative pre-mRNA splicing significantly alters the patterns of gene expression in endocrine tissues such as the thyroid or testes (11). This results in diversity of protein expression, which is important to the specific functions of differentiated endocrine tissues. In the best studied example of the calcitonin/calcitonin gene-related protein (CGRP) pre-mRNA, regulatory pre-mRNA sequence elements have been mapped, and several RNA-binding proteins have been identified, which may function to control the tissue-specific splicing patterns.

The expression of several endocrine genes can be regulated at the level of translation from mRNA into protein. This process occurs in the cytoplasm. Examples include insulin-like growth factor 2 (IGF2) and fibroblast growth factor 2 (FGF2), (12) which have alternative translation initiation sites. The use of the alternative sites is regulated by the state of cellular growth and proliferation. IGF2 has two mRNAs. A minor 4.8 kb species is translated constitutively. A major 6.0 kb variant is generally sequestered and remains untranslated in a 100S ribonucleoprotein particle. In growing cells, however, the 6.0 kb variant is mobilized and translated by the mediation of a kinase signalling pathway. RNA-binding proteins and the signalling pathways have been identified which regulate IGF2 translation (13).

The general translational apparatus and the activity of the eukaryotic initiation factors (eIFs) may also be regulated by external signals, via phosphorylation pathways. One example is eIF4E, a component of the cap-binding complex. This complex binds the m⁷G cap of mRNA to increase its interaction with the ribosome. Growth factors and hormones increase the state of eIF4E phosphorylation, and this may be associated with increases in the rate of translation of certain mRNA transcripts (14). A further level of complexity is added by the existence of a pathway for degrading mRNA in a mechanism that can be linked to the translational apparatus. This cotranslational degradation pathway involves small RNA molecules of 20–30 nucleotides in length. Such small RNA molecules are classed into two categories, small interfering RNAs and microRNAs. They are also involved in many aspects of post-transcriptional gene regulation (15).

Translation may also be regulated by localization of protein production to regions of the cell where the products are required at high concentrations. This may be accomplished by restriction of the relevant mRNA to a particular region, for example the transport of β-actin mRNA into cellular processes and growth cones. This transport process is regulated by signal transduction systems (16). While mRNA localization and regional translation has been known to be a mode of regulating embryo polarity and development, its role in other cells and tissues is beginning to emerge.

The cell cycle coordinates cellular growth and division. The core of the cell cycle machinery consists of cyclins and cyclin-dependent kinases (CDKs) (17). Cyclins bind the CDKs and direct the phosphorylation activity of the CDKs to appropriate targets, for example, members of the retinoblastoma protein (Rb) family. Many hormones affect the cell cycle machinery. For example, growth factors promote progression of the cell cycle through G1 to the restriction point, at least in part, by signals to cyclin D. Conversely, inhibitory cytokines such as transforming growth factor β (TGFβ) negatively regulate the cell cycle via several pathways, including cyclin inhibitor proteins, the Smad proteins, or by interfering with MDM2 activation of Rb function (18).

One mechanism of control of the cell cycle is by the capacity of a cell to monitor its rate of cellular biosynthesis. In the budding yeast, Saccharomyces cerevisiae, the cyclin protein CLN3 (the homologue of human cyclin D) acts as a sensor of cellular biosynthesis. Once biosynthesis has exceeded a threshold, CLN3 stimulates the cell cycle, thereby triggering a cell division. Thus, a hormone with an effect on cellular biosynthesis may also potentially function to indirectly stimulate the cell cycle via a similar mechanism (19).

The cell cycle and hormonal systems may be linked in much more a complex relationship. There is recent evidence that a core protein of the cell cycle machinery, cyclin D1, can bind and activate the oestrogen receptor (ER), to enhance ER-mediated gene transcription (20). Like other steroid receptors, activation of the oestrogen receptor usually occurs when bound to the ligand, oestrogen. However, cyclin D1 activation of the ER is independent of ligand binding, nor is an interaction with a CDK needed. Furthermore, cyclin D1 acts as a bridge between the receptor and the steroid receptor coactivator proteins (SRCs). Binding between the ER and SRCs is normally regulated by the presence of ligand, so the control of the link between the receptor and its SRC partners can be subverted by cyclin D1. Thus, the activity of the ER has two inputs: by binding to its ligand, oestrogen; and via the cell cycle core protein, cyclin D1. These data show that there are intricate coordinations between the cell cycle machinery, cellular metabolism, and signalling and gene expression. The ultimate function of this coordination between cellular machines is not known.

The process of ageing is associated with a decline in function of several hormonal systems in humans (see Chapter 10.1.1). Data from genetic studies in the worm, fruit fly and mouse suggest that the converse is also true (21): hormonal factors and signalling systems may play a role in the regulation of the ageing process (Table 1.5.3).

While ageing occurs at the level of the whole organism, apoptosis is a cellular process of programmed cell death. Some apoptosis is probably required in every tissue and organ. This may be to remove diseased cells, or to control the growth and morphology of a tissue. Apoptosis is generally tightly regulated. One example of the hormonal control of apoptosis is the effect of glucocorticoids on the involution of the thymus. This is due to the induction of apoptosis in thymocytes. There are several mechanisms for the proapoptotic effect of glucocorticoids. Glucocorticoids may exert effects at the level of gene transcription, by interfering with the function and formation of the AP1 and NF-κB transcription complexes. However, glucocorticoid stimulation also results in the sequestration of NF-κB in the cytoplasm, thereby preventing its action in the nucleus. This sequestration is due to the binding of NF-κB and masking of its nuclear localization signal by the inhibitory protein, IκB. Here, glucocorticoids function indirectly, by increasing transcription and levels of IκB family members (22). The outcome of these regulatory pathways is to influence the transcription of genes controlled by NF-κB, thereby inducing apoptosis.

Several cytokines and growth factors are cellular survival factors, reducing the likelihood of a cell undergoing apoptosis. In addition to pathways controlling expression of key genes (probably signal through the ras, raf, and mitogen activated protein (MAP) kinase pathway), signalling mechanisms that regulate the apoptosis machinery itself have been defined. In the case of cytokines and growth factors, an antiapoptotic signalling pathway involves phosphatidylinositol (3,4,5) kinase and the serine-threonine kinase, akt. Akt can directly phosphorylate and regulate the activity of a precursor of the apoptosis machinery, procaspase 9 (23).

Despite the many advances, a complete picture of the molecular mechanisms underlying hormonal regulation (encompassing all components: specificity, flexibility, and feedback) is still lacking for most systems. Study of proteins and complexes involved in signal transduction and action has shown that most signalling pathways are not linear. Instead, there are many interactions between different signalling pathways. This network of interactions is now beginning to be modelled and can, to some extent, predict and explain complex biological phenomena (24). These include persistent activation of downstream effector molecules, even when the original stimulus has been removed, and gate effects, where some levels of signals are transmitted, while other levels are not. Many of these signalling networks allow for complex positive and negative feedback controls. One example is the interaction of the phospholipase C and the ras pathways, which share protein kinase C activation in their chains (Fig. 1.5.2). Both pathways are activated at their apex by the epidermal growth factor receptor. The link via protein kinase C allows the establishment of a persistent activation of the outputs of the pathways after a threshold of stimulation is reached and then withdrawn. However, the effective modelling and prediction of signalling networks will have to account for localization of signalling effectors to specific subcellular regions by anchor and scaffold proteins. Examples of such spatial restriction of signalling networks include the Smad network, and adds additional opportunities for specificity of function and regulation.

The molecular mechanisms of the response to a hormonal signal are increasingly well understood. These mechanisms involve networks of signalling effectors, which act to regulate nearly every major molecular machine of the cell. Furthermore, molecular machines do not function independently of each other. In some examples, interaction between types of cellular machines occurs via hormonal signalling intermediates. Specificity of action, once seen as simply the presence or absence of a receptor, must now be understood in terms of networks, localization of effectors, and the structural interactions between multidomain and multiprotein complexes.