P Michael Conn & Jo Ann Janovick. American Scientist. Volume 93, Issue 4. Jul/Aug 2005.
Many diseases are caused by mutations in a person’s DNA, since these errors often change the amino acid composition of proteins encoded by genes. Scientists who study genetic diseases frequently experiment on cells that contain the mutation and produce the defective protein, which usually has altered chemical properties. For example, cells with a mutant receptor may fail to bind that receptor’s ligand, and cells with altered enzymes may lose the ability to catalyze certain chemical reactions. Physicians describe such mutations as causing a “loss of function,” and most scientists assumed that the associated proteins were fatally flawed molecules. This view is undoubtedly correct in some cases, but recent evidence challenges the universality of the assumption. According to data from many laboratories (including ours), disease-causing mutations frequently cause otherwise functional proteins to be misfolded and misrouted-packages addressed but undeliverable within the cell, like a letter with the wrong zip code and an illegible address. This idea has far-reaching implications for the treatment of disease, because correcting the way a mutant protein is routed in the cell is much easier than replacing a mutant gene with gene therapy.
Proteins are misrouted when they fold the wrong way. Such errors are common, even in normal cells, despite the presence of special molecules called chaperones, which help proteins fold into the correct three-dimensional shape, or conformation. Proteins that do not fold correctly, including most mutant proteins, are flagged by the cell’s quality-control (QC) mechanism and slated for destruction or reprocessing.
Our lab has recently experimented with pharmacological chaperones (we call them pharmacoperones) that correct the intracellular routing of a misfolded yet potentially functional protein, thereby reversing in cultured cells one cause of a hormonal disease called hypogonadotropic hypogonadism. The results suggest a general form of therapy that will, like gene therapy, treat disease by restoring a missing protein, with fewer potential risks than that controversial practice. In principle, it’s possible that scientists could apply this approach to any disease caused by misfolded proteins-a group that includes congenital disorders such as cystic fibrosis and retinitis pigmentosa and late-onset conditions such as Alzheimer’s disease, cataracts and certain types of cancer.
Supervised Molecular Origami
Protein folding is a complex process that is especially difficult among the molecules crowded inside a cell. Depending on the protein and its eventual destination, these molecules are synthesized by ribosomes in the cytoplasm or by ribosomes embedded in a specialized compartment called the endoplasmic reticulum, or ER. The protein products can fold immediately or over a period of days, floating freely in the cytoplasm or contained in the ER or other organelles. Regardless of the rate or site of synthesis, most proteins adopt their final shape in several stages. According to current models, the protein backbone folds in groups of four or more amino acids to limit interference from bulky or charge-bearing portions of the molecule. Biochemists now believe that proteins do not spontaneously fold into their final, active conformation.
In plant and animal cells, proteins destined to be secreted or embedded in the cell membrane are synthesized in the ER, where a thick soup of templates, enzymes and sugars can promote some conformations and inhibit others. These guides execute sequential rounds of intricate modifications, a process that is closely monitored by the cell’s QC system. This system supervises new proteins from synthesis to delivery within the cell, and it flags misfolded and improperly processed proteins with endogenous chaperones and sensors that recognize incorrect conformations. By helping to remove misfolded proteins, this system protects the cell from the aberrant activity of such molecules and maintains the proper balance between protein synthesis, maturation and degradation. It also prevents the accumulation of defective proteins and manages the transport, or trafficking, of completed proteins to other cell compartments. With its quality-control watchdogs, the ER can also avoid the premature exit of partially folded or incompletely assembled proteins that may be toxic to the cell.
The effectiveness of the QC system depends, in large part, on its ability to distinguish incompletely or incorrectly processed proteins. One hallmark of a correctly folded protein is the absence of exposed hydrophobic regions. Chaperones in the ER interact with such regions and retain the proteins that contain them. The QC mechanism also utilizes the addition and removal of sugar molecules to the protein. These are linked together in a pattern that helps to identify abnormal molecules.
Cell biologists initially believed that faulty proteins and protein fragments were digested locally in the endoplasmic reticulum. However, we now know that terminally misfolded molecules are sent out of the ER for degradation by proteasomes, macromolecules that chop up unwanted proteins all over the cell. The process of exporting a misfolded, chaperone-bound protein is linked to the process of recruiting a proteasome. In both cases, the covalent addition of many copies of a small protein called ubiquitin marks the protein slated for destruction. A cluster of these tags might help extract the polypeptide from the ER membrane by bringing in accessory molecules, and it also forms a beacon for the proteasome complex.
A fundamental limitation of the QC system is that it relies on chemical, not functional, tests for misfolded proteins. Although the presence of an exposed hydrophobic surface often marks misfolded or deleterious proteins, this test is unrelated to whether the protein actually works. In understanding the relation between the QC system and the pharmacoperone data, it is helpful to recognize that not all mutant proteins destined for destruction are nonfunctional.
Other factors besides mutations can trigger abnormal folding, including protein overabundance, temperature changes, oxidative stress and some cellular signals. The cell experiences aberrant folding of some proteins as a loss of function, because none of those molecules escapes destruction by the QC system. Alternatively, misfolded proteins may stick together, leading to potentially toxic aggregations in the cell or the accumulation of extracellular amyloid. In the past decade, scientists have striven to understand how abnormal folding causes certain diseases and to design therapies that prevent or correct the structural abnormalities. In this regard, the rescue of misfolded, “trafficking-defective” proteins by pharmacological chaperones is one of the most promising strategies.
Misfolding May Mean Misrouting
Our interest in protein folding began with the discovery of a mutation in the gene that encodes a receptor for gonadotropin releasing hormone (GnRH). The mutant gene came from a male patient with a disorder called hypogonadotropic hypogonadism (HH). This disease precludes the release of sufficient testosterone, preventing normal spermatogenesis. In most men, the hypothalamus releases GnRH into the bloodstream, causing the secretion of luteinizing hormone (LH) from the pituitary gland and the consequent release of testosterone from the gonads. However, people with HH usually have little or no LH, therefore scant testosterone. In this case, the patient’s LH level failed to rise when his physician, our collaborator Juan Pablo Mendez at the Mexican Institute of Social Security, gave a dose of GnRH. This result made us suspect that the fault lay with that hormone’s receptor. When members of Mendez’s lab sequenced the gene, they found a mutation that changed the 90th amino acid from a negatively charged glutamic acid to a positively charged lysine. We called the mutant E90K, using the single-letter abbreviations for the amino acids before and after the change at the 90th position.
When our team in Oregon put the cloned gene into cultured cells, those containing the mutant version did not bind GnRH or initiate the appropriate intracellular signal. We concluded that the mutant was defective in its ability to bind GnRH-a perfectly reasonable interpretation based on the information available at the time.
Then we decided to try one of those experiments that you do just to satisfy your curiosity. We made a version of the E90K receptor with modifications that directed the protein to the plasma membrane-the requisite place for proper function of a receptor. These so-called targeting sequences were far from the 90th amino acid and unlikely to interact with it. To our surprise, the E90K mutant worked fine when targeted to the plasma membrane-better, in fact, than the wild-type (unmutated) version! The immediate and inescapable conclusion was that the mutant receptor was not defective at all-just unable to get to the plasma membrane.
Pharmacoperones to the Rescue
This experiment taught us something in an academic sense but had little immediate clinical value. We thought about how to salvage a misrouted (misfolded) protein and seized on the use of peptidomimetic compounds-structures that look like peptide ligands for the receptor but are smaller and, unlike peptides, can enter cells. We decided to use antagonists, because we did not want to activate the receptor during the process of rescuing it.
We wanted a chemical structure that was small and hydrophobic-two requirements for entry into cells-and one that could serve as a template on which the mutant could fold correctly and sneak past the QC system of the cell. Our ideal molecule would bind the GnRH receptor with high specificity and high affinity. The specificity was important to minimize interactions with other proteins in the cell-interactions that would cause side effects if the agent were ever used in the clinic. A high chemical affinity between the receptor and our template would maximize the effects of the drug, but we also wanted to wash the molecule off the receptor once it was rescued, so the binding had to be reversible too. The only other requirement was that the molecule needed to be stable in the cellular environment. The data would be much easier to decipher if we didn’t have to worry about the compound falling apart during the experiment.
Colleagues Wallace T. Ashton and Mark T. Goulet at Merck and Company had developed a molecule called IN3 with just these properties. The compound was originally created in an effort to make a small drug that would enter the circulation and bind to the GnRH receptor.
To test the effect of our exogenous template, we simply added the drug to cultured cells with the mutant version of the GnRH receptor. We hoped that IN3 would enter the cells and the ER, bind the mutant protein as it was being synthesized and encourage its correct folding. If the E90K were folded the right way, it ought to be able to pass the QC apparatus and take a normal route to the membrane. Once at the surface, the receptor’s seven loops through the membrane should hold it in the correct shape, and we could remove the drug, leaving behind a functional receptor. Removing IN3 was a critical feature-as an antagonist, the drug would occupy the receptor and prevent activation by normal levels of GnRH unless we got rid of it.
It worked: The template-drug successfully rescued the mutant receptor from the QC scrap heap, and we proved that the receptor made its way to the surface of the cell in the presence of IN3. But did the mutant behave like the normal receptor? We confirmed that the rerouted mutant recognized the same agonists and antagonists as the wild-type receptor and bound them with the same affinities. The turnover of the mutant protein was the same, indicating that it was recycled normally from the plasma membrane. Furthermore, the receptor seemed to couple normally with its effector proteins, as the mutant generated similar intracellular messengers in response to the hormone as did the wild-type. In each test, the rescued version functioned just like the normal version. The mutant behaved like a bright adolescent-it never lost function, only direction.
Rescuing Other Mutations
By the time we completed these studies, other laboratories had reported 16 more mutations in GnRH receptor genes from patients with HH. We wanted to know whether the effectiveness of our technique for rescuing this receptor was limited to the E90K mutant or whether it could also work for others as well. The 17 errors were scattered throughout the gene: Fifteen of them led to single amino acid changes in the protein, one caused incorrect splicing of the messenger RNA, and another caused the receptor to be much shorter than normal. Our rescue strategy didn’t apply to the latter two, because they didn’t make a full-length protein.
We reconstructed all of the mutant genes in the lab, transfected them into cultured cells and confirmed the production of receptor protein and the cells’ inability to bind GnRH. And when we added IN3, voilá: Thirteen were rescued. Of the 15 amino-acid-substitution mutants, the two that failed lay very close to each other. One could easily imagine that changes to a critical region for ligand binding or effector activation might silence a perfectly folded and situated protein (as we used to think was the case for nearly all loss-of-function mutations). Or perhaps the two nonresponders were so drastically misfolded that IN3 could not rescue them.
We noted with interest that 11 of the 15 amino acid substitutions altered the ionic charge at that site (for example, positive to negative, or uncharged to either positive or negative). These modifications, which are minor in terms of the overall charge of the protein, appear sufficient to alter the structure of the GnRH receptor. It was remarkable to us that a single charge out of 328 amino acids could effect this change. The intimation here was that the human GnRH receptor is highly sensitive to perturbations.
None of the mutations from HH patients was conservative-that is, a change to an amino acid with similar properties, such as the replacement of alanine with glycine or threonine with serine (these pairs differ by only a single carbon atom). Similarly, there were no disease-causing examples of simple hydrophobic-for-hydrophobic exchanges (valine for alanine, for example), positive-for-positive exchanges (lysine for arginine) or negative-for-negative exchanges (aspartate for glutamate). Some of these mutations are sure to occur, but they must either be clinically silent (meaning that they do not cause disease) or, alternatively, cause the death (or infertility) of any embryo that carries them.
The Dominant Negative Effect
The movement of transmembrane proteins, including the GnRH receptor, from the endoplasmic reticulum to the plasma membrane is quite complex. Proteins that are comfortable in an oily lipid membrane are not soluble in the watery contents of the rest of the cell. In the case of a protein with seven transmembrane domains (including the GnRH receptor), the alternating stretches of hydrophilic and hydrophobic amino acids present difficulties for transport though the cytoplasm. The cell appears to address this problem by moving a large group, or oligomer, of the receptors to the plasma membrane all at once. This practice hides the hydrophobic domains from the cytosol (the fluid in the cell) by pointing them at one another and presents the hydrophilic faces outward to interact with the surrounding water, thereby easing the job of getting receptors to the plasma membrane from a thermodynamic point of view.
A negative aspect of oligomerization is that one bad apple can spoil the barrel, a scenario that plays out when a person has one normal and one bad copy of a gene. Although an oligomer might contain a mix of well-folded and misfolded proteins, the quality-control system frequently recognizes the mutant and tags the entire structure for disposal. The mutant protein is destroyed, but so is the wild-type. Geneticists describe this occurrence as a “dominant negative” effect-the mutation actively damages something in a way that the mere absence of the same gene would not.
Using confocal microscopy, we visualized this phenomenon by fusing the gene for the GnRH receptor to the gene that encodes green fluorescent protein (GFP). The so-called chimeric protein that resulted showed the location of GnRH receptors inside the cell by glowing under a certain wavelength of light. In these experiments, cells containing the chimeric receptor showed some of the green signal in the ER (which is counterstained blue), but most of it was properly routed to the outer membrane. However, when we added the E90K mutant at the same time as the chimera, virtually all the fluorescent signal was locked up in the endoplasmic reticulum, unable to reach the surface. The pharmacoperone IN3 reversed this pattern, restoring the chimera to its proper place on the outer membrane even in the presence of the mutant receptor. Thus, IN3 rescued not only the mutant version, but the normal one as well.
Rules for Rescue
The use of pharmacoperones for protein rescue has the potential to treat any disease caused by a misfolded or misrouted protein. But what general lessons from our proof-of-principle experiments with HH apply to therapies for protein-folding diseases such as Parkinson’s, Creutzfeldt-Jacob (the human version of mad cow disease) or hypertrophic cardiomyopathy?
To see whether our results were unique to the IN3 compound, we looked for other molecules that might also rescue GnRH receptor mutants. A closely related compound with a very low binding affinity did not work at all, and protein antagonists also failed despite their binding of wild-type receptors that were on the surface of the cultured cells. We concluded that the proteins were either too large or too charged to enter the cell and bind the receptor.
However, we did succeed with other antagonists that, like IN3, mimicked a peptide. Several indoles, quinolones and erythromycin macrolides made the grade. The human eye would never recognize these compounds as things that bind to the GnRH receptor, but the receptor does. All of them rescued E90K to varying degrees, and we used this information to create “rules for rescue.” Briefly, molecules within a chemical group were effective in proportion to their affinity for the GnRH receptor-high-affinity ligands within each group worked best. Between groups, the relative solubility-the ability to permeate the cell membrane and reach the nascent protein-was most important.
One could also imagine that a pharmacoperone could stabilize a protein without binding to the same site as a natural ligand. Such a characteristic might actually be advantageous, because clinicians might not have to remove this drug before the rescued mutant could work. This idea made us think of all the thousands of test compounds that drug companies screen for specific properties. They get a hit when some molecule acts as an agonist or antagonist for their target of interest. By using only this strategy, industry chemists would miss a pharmacoperone that was neither an agonist nor an antagonist. The inescapable conclusion of this syllogism is that pharmaceutical archives may contain many useful structures that escaped detection because the drug companies used assay criteria that were too narrow.
Although rescuing mutant proteins by helping them fold may become a significant form of therapy, one could also imagine drugs that applied the same approach to normal proteins. If a pharmacoperone altered the folding kinetics of a normal protein, it could cause misfolding and degradation. Such an approach, which we call “protein shipwrecking,” might be an effective way to inhibit out-of-control cancer genes, or it might provide a novel approach to contraception.
Considering the Next Target
The data from our experiments with the GnRH receptor allowed us to make some predictions about which disease-causing mutants would be good candidates for rescue with small, permeable peptidomimetics. First, the abnormal protein should not bear mutations in amino acids that are essential for ligand binding (for receptor proteins), substrate or cofactor binding (for enzymes), ion binding (for ion channels) or interaction with other effectors (G-proteins and other signaling molecules). Such mutations block the function of even a properly folded and targeted mutant. second, large deletions or truncations yield an irredeemable protein-the parts needed to reestablish higher-order structures do not exist.
Third, the loss or gain of certain amino acids may represent too great an obstacle for pharmacological rescue. For example, the amino acid cysteine forms covalent bonds with other cysteines in the protein, bridges that are important for a protein to hold its shape. The absence or inappropriate formation of such connections may be so disruptive that the protein cannot fold correctly even with a template. Similarly, the loss or gain of the amino acid proline may limit or block mutant rescue because it forces a turn in the protein backbone. In some proteins, an abrupt turn is probably required for the native structure in a way that pharmacoperones cannot correct. Other dubious candidates for rescue are those mutants in which amino acid substitutions impede or promote hydrogen-bond formation, a process that often establishes the necessary interactions between different portions of a complex protein.
Fourth, mutations that substitute smaller residues (glycine or alanine) are probably favorable candidates for pharmacoperone therapies. In contrast to the problems of swapping in a large amino acid (such as valine, tryptophan or threonine), smaller replacements allow more freedom of movement, thus greater tolerance for template-mediated folding and a (potentially) more efficient pharmacological rescue.
Misfolding as a Disease Etiology
In addition to mutations in specific genes, all kinds of cellular stresses can cause widespread faults in the folding, modification or routing of proteins, any of which can mean their degradation at the hands of the quality-control machinery. However, too many defective molecules in a cell could conceivably clog the system and cause an accumulation of imperfect proteins in the ER. Many scientists now believe that the accumulation and aggregation of misfolded proteins are responsible for neurodegenerative diseases such as early-onset Alzheimer’s, Parkinson’s and prion disease, as well as early-onset cataracts, α1-antitrypsin deficiency, type II diabetes mellitus and systemic amyloidosis. In all of these diseases, proteins or protein fragments convert from normal, soluble conformations to insoluble, sticky fibers known as amyloids. Typified by β-amyloid in Alzheimer’s disease and α-synuclein in Parkinson’s disease, these proteins coalesce into fibrillar aggregates that have a characteristic structure. The insoluble dumps can form either inside or outside the cell.
Unlike the aggregate-forming amyloids, most other protein-folding diseases simply result from impaired intracellular trafficking or degradation of the mutant protein. Many such disorders involve membrane-associated proteins, including retinitis pigmentosa, cystic fibrosis, diabetes insipidus and some forms of familial hypercholesterolemia. For the latter, several mutations in the low-density lipoprotein receptor cause defective processing that attracts attention from the QC system, which then targets the receptor for degradation. Cystic fibrosis is caused by mutations in the CFTR gene, which codes for a channel that conducts chloride ions through the cell’s outer membrane. The most common mutation is a deletion of the 508th amino acid, which leads to chaperone-mediated ER retention and rapid proteolysis of the incompletely processed (albeit functional) CFTR protein. Thus, the channel cannot get to the plasma membrane, and the cell cannot regulate the flow of chloride ions. Similar mechanisms can explain cases of nephrogenic diabetes insipidus (mutations in the gene that encodes the aquaporin-2 water channel) and retinitis pigmentosa (mutations in the gene encoding rhodopsin).
With the number of human and animal diseases caused by misfolded proteins, many academic and industry scientists are pursuing therapies that address this central pathogenic mechanism. Pharmacoperones, small molecules that easily penetrate the cell, bind to the mutant protein and encourage it to fold correctly, seem to be widely applicable to the spectrum of protein-folding diseases, and we and others believe they have distinct advantages over some other therapeutic approaches. Clinical trials that use pharmacoperones to treat hypogonadotropic hypogonadism, cystic fibrosis and α1-antitrypsin deficiency are soon to come, and other compounds with similar attributes are very likely to figure in the future of drug development.