Proteins have the most dynamic and diverse role of any macromolecule in the body, catalysing biochemical reactions, forming receptors and channels in membranes, providing intracellular and extracellular scaffolding support, and transporting molecules within a cell or from one organ to another. It is currently estimated that there are 25,000–40,000 different genes in the human genome, and with alternative splicing of genes and post-translational modification of proteins (for example, by cleavage, phosphorylation, acylation and glycosylation), the number of functionally distinct proteins is likely to be much higher1,2,3. Viewed from the perspective of disease mechanisms, these estimates pose an immense challenge to modern medicine, as disease may result when any one of these proteins contains mutations or other abnormalities, or is present in an abnormally high or low concentration. Viewed from the perspective of therapeutics, however, these estimates represent a tremendous opportunity in terms of harnessing protein therapeutics to alleviate disease. At present, more than 130 different proteins or peptides are approved for clinical use by the US Food and Drug Administration (FDA), and many more are in development.

Protein therapeutics have several advantages over small-molecule drugs. First, proteins often serve a highly specific and complex set of functions that cannot be mimicked by simple chemical compounds. Second, because the action of proteins is highly specific, there is often less potential for protein therapeutics to interfere with normal biological processes and cause adverse effects. Third, because the body naturally produces many of the proteins that are used as therapeutics, these agents are often well tolerated and are less likely to elicit immune responses. Fourth, for diseases in which a gene is mutated or deleted, protein therapeutics can provide effective replacement treatment without the need for gene therapy, which is not currently available for most genetic disorders. Fifth, the clinical development and FDA approval time of protein therapeutics may be faster than that of small-molecule drugs. A study published in 2003 showed that the average clinical development and approval time was more than 1 year faster for 33 protein therapeutics approved between 1980 and 2002 than for 294 small-molecule drugs approved during the same time period4. Last, because proteins are unique in form and function, companies are able to obtain far-reaching patent protection for protein therapeutics. The last two advantages make proteins attractive from a financial perspective compared with small-molecule drugs.

A relatively small number of protein therapeutics are purified from their native source, such as pancreatic enzymes from hog and pig pancreas5,6 and α-1-proteinase inhibitor from pooled human plasma7,8, but most are now produced by recombinant DNA technology and purified from a wide range of organisms. Production systems for recombinant proteins include bacteria, yeast, insect cells, mammalian cells, and transgenic animals and plants9,10,11,12,13. The system of choice can be dictated by the cost of production or the modifications of the protein (for example, glycosylation, phosphorylation or proteolytic cleavage) that are required for biological activity. For example, bacteria do not perform glycosylation reactions, and each of the other biological systems listed above produces a different type or pattern of glycosylation. Protein glycosylation patterns can have a dramatic effect on the activity, half-life and immunogenicity of the recombinant protein in the body. For example, the half-life of native erythropoietin, a growth factor important in erythrocyte production (see below), can be lengthened by increasing the glycosylation of the protein. Darbepoetin-a is an erythropoietin analogue that is engineered to contain two additional amino acids that are substrates for N-linked glycosylation reactions. When expressed in Chinese hamster ovary cells, the analogue is synthesized with five rather than three N-linked carbohydrate chains; this modification causes the half-life of darbepoetin to be threefold longer than that of erythropoietin14.

Perhaps the best example of trends in the production and use of protein therapeutics is provided by the history of insulin in the treatment of diabetes mellitus type I (DM-I) and type II (DM-II). Untreated, DM-I is a disease that leads to severe wasting and death due to lack of the protein hormone insulin, which signals cells to perform numerous functions related to glucose homeostasis and intermediary metabolism15. In 1922, insulin was first purified from bovine and porcine pancreas and used as a life-saving daily injection for patients with DM-I16. At least three problems hindered the widespread use of this protein therapy: first, the availability of animal pancreases for purification of insulin; second, the cost of insulin purification from animal pancreas; and third, the immunological reaction of some patients to animal insulin. These problems were addressed by isolating the human insulin gene and engineering Escherichia coli to express human insulin by using recombinant DNA technology. By growing vast quantities of these bacteria, large-scale production of human insulin was achieved. The resulting insulin was abundant, inexpensive, of low immunogenicity and free from other animal pancreatic substances. Recombinant insulin, approved by the US FDA in 1982, was the first commercially available recombinant protein therapeutic, and has been the major therapy for DM-I (and a major therapy for DM-II) ever since16,17,18,19,20.

Recombinantly produced proteins can have several further benefits compared with non-recombinant proteins. First, transcription and translation of an exact human gene can lead to a higher specific activity of the protein and a decreased chance of immunological rejection. Second, recombinant proteins are often produced more efficiently and inexpensively, and in potentially limitless quantity. One striking example is found in the protein-based therapy for Gaucher's disease, a chronic congenital disorder of lipid metabolism caused by a deficiency of the enzyme β-glucocerebrosidase (also known as glucosylceramidase) that is characterized by an enlarged liver and spleen, increased skin pigmentation and painful bone lesions21,22. At first, β-glucocerebrosidase purified from human placenta was used to treat this disease, but this requires purification of protein from 50,000 placentas per patient per year, which obviously places a practical limit on the amount of purified protein available. A recombinant form of β-glucocerebrosidase was subsequently developed and introduced, which is not only available in sufficient quantities to treat many more patients with the disease, but also eliminates the risk of transmissible (for example, viral or prion) diseases associated with purifying the protein from human placentas23,24,25. This also illustrates a third benefit of recombinant proteins over non-recombinant proteins — the reduction of exposure to animal or human diseases.

A fourth advantage is that recombinant technology allows the modification of a protein or the selection of a particular gene variant to improve function or specificity. Again, recombinant β-glucocerebrosidase provides an interesting example. When this protein is made recombinantly, a change of amino-acid arginine-495 to histidine allows the addition of mannose residues to the protein. The mannose is recognized by endocytic carbohydrate receptors on macrophages and many other cell types, allowing the enzyme to enter these cells more efficiently and to cleave the intracellular lipid that has accumulated in pathological amounts, which results in an improved therapeutic outcome23. Last, recombinant technology allows the production of proteins that provide a novel function or activity, as discussed below.

The 25 years since the approval of recombinant insulin by the FDA have seen a remarkable expansion in the number of therapeutic applications of proteins. More than 130 proteins (over 95 of which are produced recombinantly) are currently approved for clinical use by the FDA, and many more are in development. An appreciation of the many therapeutic uses of proteins may be facilitated by categorizing such therapies according to their mechanism of action, and, in this article, we summarize currently approved protein therapeutics by suggesting a classification system that is based on their pharmacological action (Box 1). Examples of protein therapeutics in each category and clinical conditions in which they are used are discussed in the text, and a listing of FDA-approved protein therapies and their functions and clinical uses is presented in Tables 1, 2, 3, 4, 5, 6, 7, 8. Examples of protein-based vaccines and diagnostics that highlight the growing importance of proteins in medicine are provided in Tables 9, 10.

Table 1 Protein therapeutics replacing a protein that is deficient or abnormal (Group Ia)*
Table 2 Protein therapeutics replacing a protein that is deficient or abnormal (Group Ia)*
Table 3 Protein therapeutics augmenting an existing pathway (Group Ib)*
Table 4 Protein therapeutics augmenting an existing pathway (Group Ib)*
Table 5 Protein therapeutics providing a novel function or activity (Group Ic)
Table 6 Protein therapeutics that interfere with a molecule or organism (Group IIa)*
Table 7 Protein therapeutics that interfere with a molecule or organism (Group IIa)*
Table 8 Protein therapeutics that deliver other compounds or proteins (Group IIb)
Table 9 Protein vaccines (Group III)*
Table 10 Protein diagnostics (Group IV)

Group I: enzymes and regulatory proteins

Protein therapeutics in this group function by a classic paradigm in which a specific endogenous protein is deficient, and the deficit is then remedied by treatment with exogenous protein. Protein therapeutics that we have classified in Group Ia are used to replace a particular activity in cases of protein deficiency or abnormal protein production. These proteins are used in a range of conditions, from providing lactase in patients lacking this gastrointestinal enzyme26 to replacing vital blood-clotting factors such as factor VIII27,28 and factor IX29,30 in haemophiliacs. A classic example, as mentioned above, is the use of insulin for the treatment of diabetes. Another important example is in the treatment of cystic fibrosis, a common lethal genetic disorder. In this disease, defects in the chloride channel encoded by the CFTR gene lead to abnormally thick secretions, which can (among other effects) block pancreatic enzymes from travelling down the pancreatic duct into the duodenum31. This prevents food from being properly digested and results in malnutrition. Patients with cystic fibrosis are often treated with a combination of pancreatic enzymes isolated from pigs — including lipases, amylases and proteases — that allow the digestion of lipids, sugars and proteins. Patients who have had their pancreas removed or who suffer from chronic pancreatitis can also benefit from this therapy5,6. Other striking examples include various diseases caused by metabolic enzyme deficiencies, such as Gaucher's disease as mentioned above, mucopolysaccharidosis, Fabry disease and others. Additional protein therapies that replace a particular activity are listed in Tables 1, 2.

It may sometimes be desirable to enhance the magnitude or timing of a particular normal protein activity, and protein therapeutics that we have classified in Group Ib are administered to achieve this. Such protein therapeutics have been successful in treating haematopoietic defects; the most prominent example is recombinant erythropoietin, a protein hormone secreted by the kidney that stimulates erythrocyte production in the bone marrow31. In patients with chemotherapy-induced anaemia or myelodysplastic syndrome, recombinant erythropoietin is used to increase erythrocyte production and thereby ameliorate the anaemia. In patients with renal failure, whose levels of endogenous erythropoietin are below normal, recombinant protein is administered to correct this deficiency32,33,34,35,36. Another example is provided by the treatment of neutropaenic patients with granulocyte- or granulocyte-monocyte colony stimulating factor (G-CSF or GM-CSF, respectively)36,37, which stimulate an increase in the number of neutrophils produced by the bone marrow to allow these patients to better combat microbial infections. Similarly, thrombocytopaenic patients can be treated with interleukin 11 (IL11)38, which increases platelet production and thereby prevents bleeding complications.

In vitro fertilization (IVF) is another area in which Group Ib proteins are applied. Increased levels of follicle-stimulating hormone (FSH) are normally produced by the anterior pituitary gland just before ovulation. These high levels of FSH can be enhanced by treatment with recombinant FSH, leading to maturation of an increased number of follicles and to an increased number of oocytes available for IVF39,40. Similarly, recombinant human chorionic gonadotropin (HCG)41 is used in assisted reproductive technology to promote follicle rupture, a process that must occur before the oocytes can be transported into the fallopian tubes for fertilization.

Group Ib proteins can also have life-saving effects on thrombosis and haemostasis. Alteplase (recombinant tissue plasminogen activator (tPA; also known as PLAT)), is used to treat life-threatening blood clots in conditions such as coronary artery occlusion, acute ischaemic stroke and pulmonary embolism42,43,44,45,46. Endogenous tPA is secreted by the endothelial cells that line blood vessels. The secreted tPA normally cleaves plasminogen to plasmin, which then degrades fibrin and thereby lyses fibrin-based clots15. Although endogenous tPA may be present at normal or even increased levels near the site of a blood clot, administration of relatively large amounts of exogenous tPA may be required to disrupt these clots. Reteplase, a genetically modified form of recombinant tPA, is used to treat acute myocardial infarction47,48, and enecteplase, another genetically engineered derivative of tPA, has greater specificity than tPA for binding to plasminogen and therefore causes a more efficacious lysis of fibrin in blood clots49,50. Supraphysiological levels of coagulation factor VIIa may catalyse thrombosis and thereby stop life-threatening bleeding in patients with haemophilia A or B51,52. Also, recent studies have suggested that recombinant activated protein C53,54 can improve immunoregulation and prevent excessive clotting reactions in patients with severe, life-threatening sepsis and organ dysfunction. Many other Group Ib protein therapeutics are also used for immunoregulation — chronic hepatitis B and C, Kaposi's sarcoma, melanoma, and some types of leukaemia and lymphoma have been treated with various forms of interferon, as noted in Table 3. Other disease states treated with Group Ib proteins are summarized in Tables 3, 4.

Occasionally, the activity of a particular protein is desirable even though the body does not normally express that activity. Protein therapeutics that we have classified in Group Ic contain examples of this paradigm, including foreign proteins with novel functions and endogenous proteins that act at a novel time or place in the body. Papain, for example, is a protease purified from the Carica papaya fruit. This protein is used therapeutically to degrade proteinaceous debris in wounds55. Collagenase, obtained from fermentation by Clostridium histolyticum, can be used to digest collagen in the necrotic base of wounds56,57. The protease-mediated debridement or removal of necrotic tissue is useful in the treatment of burns, pressure ulcers, post-operative wounds, carbuncles and other types of wounds. Recombinant human deoxyribonuclease I (DNASE1) also has an interesting novel use. Normally found inside human cells, this recombinant enzyme can be used to degrade the DNA left over from dying neutrophils in the respiratory tract of patients with cystic fibrosis58. Such DNA could otherwise form mucus plugs that obstruct the respiratory tract and lead to pulmonary fibrosis, bronchiectasis and recurrent pneumonias. Thus, recombinant protein technology has allowed the therapeutic application of a normally intracellular enzyme in a novel extracellular environment.

There are many other successful examples of this approach to protein therapy. For instance, certain forms of acute lymphoblastic leukaemia are unable to synthesize asparagine and therefore require the availability of this amino acid to survive. L-Asparaginase, purified from E. coli, can be used to lower serum levels of asparagine in such patients and thereby inhibit cancer cell growth59,60. Studies of the medical leech, Hirudo medicinalis, revealed that its salivary gland produces hirudin, a potent thrombin inhibitor. The gene for this protein was then identified, cloned and used recombinantly to provide a new protein therapy, lepirudin, which prevents clot formation in patients with heparin-induced thrombocytopaenia61,62. Other organisms can also be used to produce proteins that are capable of breaking up clots that have already formed; for example, streptokinase is a plasminogen activating protein produced by group C β-haemolytic streptococci63,64,65,66. Many more therapeutic proteins that provide a novel function or activity are presented in Table 5.

Group II: targeted proteins

The exquisite binding specificity of monoclonal antibodies and immunoadhesins67 can be exploited in numerous ways using recombinant DNA technology. Many protein therapeutics that we have classified in Group IIa use the antigen recognition sites of immunoglobulin (Ig) molecules or the receptor-binding domains of native protein ligands to guide the immune system to destroy specifically targeted molecules or cells. Other monoclonal antibodies and immunoadhesins neutralize molecules by simple physical occupation of a functionally important region of the molecule. Immunoadhesins combine the receptor-binding domains of protein ligands with the Fc region of an Ig. The Fc region can target a soluble molecule for destruction because cells of the immune system can recognize the Fc region, endocytose the attached molecule and break down the molecule chemically and enzymatically. When bound to specifically recognized molecules on the surface of a cell, the Fc region can target the cell for destruction by the immune system. Cell killing can be mediated by macrophages, by other immune cells or by complement fixation.

Several Group IIa protein therapeutics have been approved for the treatment of inflammatory diseases, such as the immunoadhesin etanercept, which is a fusion between two human proteins: tumour necrosis factor (TNF) receptor and the Fc region of the human antibody protein IgG1. The TNF receptor portion of the molecule binds excess TNF in the plasma, while the Fc portion of the molecule targets the TNF for destruction. By combining these two functions, the drug neutralizes the deleterious effects of TNF (a cytokine that stimulates increased activity of the immune system) and thereby provides an effective therapy for inflammatory arthritis and psoriasis68,69,70. Another Group IIa protein that targets TNF is infliximab. This recombinantly produced monoclonal antibody binds to TNFα, and is used to neutralize the action of TNFα in inflammatory conditions such as rheumatoid arthritis and inflammatory bowel disease71,72,73.

Some Group IIa proteins are used to treat infectious diseases. Patients at high-risk for severe respiratory syncytial virus (RSV) infection, one of the leading causes of hospital admissions for paediatric respiratory illness, are given a recombinant monoclonal antibody, palivizumab, which binds to the RSV F protein and thereby directs the immune-mediated clearance of the virus from the body74,75. Enfuvirtide is an example of a Group II protein therapeutic that is not a monoclonal antibody or an immunoadhesin. By binding to gp120/gp41 — the HIV envelope protein responsible for fusion of the virus with host cells — this 36-amino-acid peptide prevents the conformational change in gp41 that is required for viral fusion, and thereby inhibits viral entry into the cell76,77,78.

Another area in which Group IIa antibodies have been successful is oncology. For example, rituximab is a human/mouse chimeric monoclonal antibody that binds to CD20, a transmembrane protein expressed on >90% of B-cell non-Hodgkin's lymphomas, and targets the cells for destruction by the body's immune system79,80,81. Although rituximab is most often used in combination with anthracycline-based chemotherapy, it is one of the few monoclonal antibody anticancer therapies that is approved as a monotherapy. Cetuximab is a monoclonal antibody that is used to treat colorectal cancer and head and neck cancer; this monoclonal antibody binds epidermal growth factor receptor (EGFR) and impairs cancer cell growth and proliferation82. Other recently developed Group IIa protein therapeutics are listed in Tables 6, 7 and many more protein therapeutics utilizing the exquisite specificity of monoclonal antibodies are in development, especially for cancer and inflammatory diseases.

Many important processes are modulated by cell-surface receptors that are activated upon binding of their cognate ligands15. By binding to such receptors, targeted protein therapeutics may activate cell signalling pathways and profoundly affect cell function. Outcomes may range from cell death (through the induction of apoptosis), to downregulation of cell division to increased cell proliferation. Although it has been difficult to prove that a particular target-binding protein mediates an in vivo effect through the modulation of a particular signalling pathway, in vitro evidence suggests that this type of modulation is involved in the mechanism of action of certain therapeutic proteins. For example, the treatment of certain breast cancers, in which the malignant cells express the HER2/Neu (also known as ERBB2) cell surface receptor, is enhanced by the addition of trastuzumab (an anti-HER2/Neu monoclonal antibody) to the therapeutic regimen83. Although trastuzumab contains an Fc region that facilitates antibody-dependent cellular cytotoxicity mediated by natural killer cells, it seems unlikely that this is trastuzumab's only mechanism of action. Other monoclonal antibodies, with similar Fc regions and abilities to target breast cancer cells, have failed to show efficacy in vivo. Trastuzumab, however, has been shown in vitro to induce intracellular signalling events that control the growth of breast cancer cells. It is therefore likely that a combination of mechanisms accounts for the therapeutic activity of trastuzumab, including inhibition of the phosphatidylinositol 3-kinase (PI3K) pathway, inhibition of angiogenesis and inhibition of HER2 receptor cleavage84,85. The complex action of trastuzumab highlights the fact that, while modulation of cell physiology through simple receptor binding may play a role in the activity of some targeted therapies, the relative contribution of receptor binding to the overall efficacy of the therapeutic may be difficult to dissect.

One of the great challenges in drug therapy is the selective delivery of small-molecule drugs and proteins to the intended therapeutic target. The body normally uses proteins to achieve specialized transport and delivery of molecules. An active area of current research is focused on understanding the principles of protein-based, targeted delivery of molecules, so that these principles can be applied to modern pharmacotherapy. This strategy is exploited by protein therapeutics that we have classified in Group IIb (Table 8), such as gemtuzumab ozogamicin, which links the binding region of a monoclonal antibody directed against CD33 with calicheamicin, a small-molecule chemotherapeutic agent. By using this therapy, the toxic compound is selectively delivered to CD33-positive acute myeloid leukaemia cells, resulting in the selective killing of these cells86,87. Similarly, refractory CD20-positive non-Hodgkin's lymphoma cells can be destroyed selectively by ibritumomab tiuxetan, a monoclonal antibody that is directed against CD20 and linked to a radioactive yttrium isotope (Y-90)88. Another example is provided by denileukin diftitox, which uses a monoclonal antibody that is directed against the CD25 component of the IL2 receptor to deliver cytocidal diphtheria toxin to T-cell lymphoma cells that express this receptor89,90.

In addition to these current examples, interesting developments are in progress that illustrate where the field might be heading. For example, herpes simplex virus produces a protein, VP22, which enters human cells. VP22 has been used in vitro to deliver proteins or other compounds to the nucleus. In one application, VP22 was used to target the tumour suppressor protein p53 to cultured osteosarcoma cells that lacked the p53 gene (and hence the protein)91. Reintroduction of p53 led to apoptosis of the cells. It is thought that a novel and effective therapy for certain forms of cancer could use protein-based targeting of the p53 gene. Another area of research involves the delivery of proteins and other macromolecules to the CNS, which is challenging owing to the highly selective blood–brain barrier (BBB). Animal experiments have demonstrated, however, that fusion proteins combining a therapeutic protein with a protein that naturally has specific access through the BBB can allow successful delivery of the therapeutic protein to the CNS. For example, a fragment of the tetanus toxin protein that naturally crosses the BBB has been shown in animal experiments to deliver the enzyme superoxide dismutase (SOD) to the CNS92. This type of therapeutic could potentially be used to treat neurological disorders such as amyotrophic lateral sclerosis, in which CNS levels of SOD are reported to be low. Exciting prospects also exist for the treatment of other disorders of the CNS in which levels of a particular protein are abnormal.

Group III: protein vaccines

As recombinant DNA technology was being developed, great strides were also being made in understanding the molecular mechanisms that allow the immune system to protect the body against infectious diseases and cancer. Armed with this new understanding, proteins that we have classified in Group III have been successfully applied as prophylactic or therapeutic vaccines. Table 9 provides selected examples.

For humans to develop effective immunity against foreign organisms or cancer cells, immune cells such as helper T cells must be activated. Immune-cell activation is mediated by antigen-presenting cells, which display on their surface specific oligopeptides that are derived from proteins found in foreign organisms or cancer cells. Vaccination against certain organisms such as polio or measles has most often been achieved by injecting heat-killed or attenuated forms of these pathogens. Unfortunately, these methods have involved a certain amount of unavoidable risk of infection or adverse reaction. By specifically injecting the appropriate immunogenic (but non-pathogenic) protein components of a microorganism, vaccines can hopefully be created that provide immunity in an individual without exposing the individual to the risks of infection or toxic reaction.

Proteins that we have classified in Group IIIa are used to generate protection against infectious diseases or toxins. One successful example is the hepatitis B vaccine93,94. This vaccine was created by producing recombinant hepatitis B surface antigen (HBsAg) protein, a non-infectious protein of the hepatitis B virus. When immunocompetent humans are challenged and rechallenged with this protein, significant immunity results in the large majority of individuals. Similarly, the non-infectious lipoprotein on the outer surface of Borrelia burgdorferi has been engineered into a vaccine for Lyme disease (OspA)95,96. A recently approved vaccine against human papillomavirus (HPV) combines the major capsid proteins from four HPV strains that commonly cause genital warts (strains 6 and 11) and cervical cancer (strains 16 and 18)97.

In addition to generating protection against foreign invaders, recombinant proteins can induce protection against an overactive immune system that attacks its own body or 'self'. One theory is that administration of large amounts of this self-protein causes the body's immune system to develop tolerance to that protein by eliminating or deactivating cells that react against the self-protein. Proteins that we have classified in Group IIIb are used to treat patients with disorders that arise from this type of autoimmune phenomenon. Immunological acceptance of a fetus during pregnancy represents a special situation with respect to vaccine use. Occasionally, a pregnant woman can reject a fetus after she has been immunized against certain antigens carried by a fetus from a previous pregnancy. Administration of an anti-Rhesus D antigen Ig prevents the sensitization of an Rh-negative mother at the time of delivery of an Rh-positive neonate. Because the woman fails to develop antibodies directed against the fetal Rh antigens, immune reactions and pregnancy loss do not occur in subsequent pregnancies, even when the new fetus carries the Rh antigens98.

Proteins that we have classified in Group IIIc could be used as therapeutic anticancer vaccines. Although there are currently no FDA-approved recombinant anticancer vaccines, there are promising clinical trials that use patient-specific cancer vaccines. For example, a vaccine for B-cell non-Hodgkin's lymphoma uses transgenic tobacco plants (Nicotiana benthamiana)99. Each patient with this type of lymphoma has a malignant proliferation of an antibody-producing B-cell that displays a unique antibody on its surface. By subcloning the idiotype region of this tumour-specific antibody and expressing the region recombinantly in tobacco plants, a tumour-specific antigen is produced that can be used to vaccinate a patient. This process requires only 6–8 weeks from biopsy of the lymphoma to a ready-to-use, patient-specific vaccine. As the genomes of infectious organisms and the nature of autoimmune diseases and cancer are more fully elucidated, more recombinant proteins will undoubtedly be developed for use as vaccines.

Group IV: protein diagnostics

Proteins that we have classified in Group IV are not used to treat disease, but purified and recombinant proteins used for medical diagnostics (both in vivo and in vitro) are mentioned here because they are invaluable in the decision-making process that precedes the treatment and management of many diseases. Table 10 provides selected examples.

A classic example of an in vivo diagnostic is the purified protein derivative (PPD) test, which determines whether an individual has been exposed to antigens from Mycobacterium tuberculosis. In this example, a non-infectious protein component of the organism is injected under the skin of an immunocompetent individual100,101,102. An active immune reaction is interpreted as evidence that the patient has been previously infected by M. tuberculosis or exposed to the antigens of this organism.

Several stimulatory protein hormones are used to diagnose endocrine disorders. Growth hormone releasing hormone (GHRH) stimulates somatotroph cells of the anterior pituitary gland to secrete growth hormone. Used as a diagnostic, GHRH can help to determine whether pituitary growth hormone secretion is defective in patients with clinical signs of growth hormone deficiency103,104. Similarly, the recombinant human protein secretin is used to stimulate pancreatic secretions and gastrin release, and thereby aid in the diagnosis of pancreatic exocrine dysfunction or gastrinoma. In patients with a history of thyroid cancer, recombinant thyroid stimulating hormone (TSH) is an important component of the surveillance methods used to detect residual thyroid cancer cells. Before the advent of recombinant TSH, patients with a history of thyroid cancer were required to stop taking replacement thyroid hormone in order to develop a hypothyroid state to which the anterior pituitary would respond by releasing endogenous TSH. TSH-stimulated cancer cells could then be detected by radioactive iodine uptake. Unfortunately, this method required patients to experience the adverse consequences of hypothyroidism. Use of recombinant instead of endogenous TSH not only allowed patients to remain on replacement thyroid hormone but also resulted in the improved detection of residual thyroid cancer cells105,106.

Imaging agents are a broad group of protein diagnostics that can be used to help identify the presence or localization of a pathological condition. For example, apcitide is a technetium-labelled synthetic peptide that binds glycoprotein IIb/IIIa receptors on activated platelets and is used to image acute venous thrombosis107. Caromab pendetide is an indium-111-labelled anti-PSA (prostate-specific antigen) antibody that can be used to detect prostate cancer108. Protein-based imaging agents are often used to detect otherwise hidden disease so it can be treated early when treatment is most likely to succeed. Imaging agents are currently used to detect cancer, image myocardial injury or identify sites of occult infection; these agents are presented in more detail in Table 10.

There are numerous in vitro protein diagnostics and two are presented here as examples of a much larger class. Natural and recombinant HIV antigens are essential components of common screening (enzyme immunoassay) and confirmatory (western blot) tests for HIV infection. In these tests, the antigens serve as 'bait' for specific antibodies to HIV gag, pol and env gene products that have been elicited in the course of infection109,110,111. Oral versions of HIV tests have also become available. Hepatitis C infection is diagnosed by using recombinant hepatitis C antigens to detect antibodies directed against this virus in the serum of potentially infected patients112,113.

Challenges for protein therapeutics

There are now many examples in which proteins have been used successfully therapeutically. Nonetheless, potential protein therapies that have failed far outnumber the successes so far, in part owing to a number of challenges that are faced in the development and use of protein therapeutics.

First, protein solubility, route of administration, distribution and stability are all factors that can hinder the successful application of a protein therapy114,115. Proteins are large molecules with both hydrophilic and hydrophobic properties that can make entry into cells and other compartments of the body difficult, and the half-life of a therapeutic protein can be drastically affected by proteases, protein-modifying chemicals or other clearance mechanisms. One example of how such challenges are being addressed is through the production of PEGylated versions of therapeutic proteins. For example, PEG-interferon is a modified form of interferon in which the polymer polyethylene glycol (PEG) is added to prolong the absorption, decrease the renal clearance, retard the enzymatic degradation, increase the elimination half-life and reduce the immunogenicity of interferon15.

A second important challenge is that the body may mount an immune response against the therapeutic protein116. In some cases, this immune response can neutralize the protein and can even cause a harmful reaction in the patient. For example, immune responses can be generated against Group Ia therapeutic proteins used to replace a factor that has been missing since birth, as illustrated by the development of antifactor VIII antibodies (inhibitors) in patients with severe haemophilia A who are treated with recombinant human factor VIII117,118. More commonly, however, immune responses are generated against proteins of non-human origin. Until quite recently, the widespread clinical application of monoclonal antibodies had been limited by the rapid induction of immune responses against this class of therapeutic proteins. The need for antibody therapeutics that evade immune surveillance and response has been a driving force in the maturation of antibody production technology. Recombinant technology and other advances have allowed the development of various antibody products that are less likely to provoke an immune response than unmodified murine antibodies. In humanized antibodies, portions of the antibody that are not critical for antigen-binding specificity are replaced with human Ig sequences that confer stability and biological activity on the protein but do not provoke an anti-antibody response; and fully human antibodies can be produced using transgenic animals or phage display technologies67,119.

The field of cancer therapeutics illustrates the pace of advances in monoclonal antibody development. In the 1980s, most of the monoclonal cancer therapeutics were murine, although there were a few examples of chimeric antibodies and isolated instances of humanized and human antibodies in clinical development. During the 1990s, humanized and fully human antibodies became the most common types of antibodies introduced into clinical trials. Since 2000, there has been a further increase in the proportion of antibodies that are fully human, with the proportion of murine and chimeric antibodies being introduced into clinical trials decreasing accordingly120.

More heavily engineered protein therapies that are based on human antibodies have also been developed over the past 10–15 years. One example is the 'minibody' AMG 531, which is currently in clinical trials for the treatment of immune thrombocytopaenic purpura. This construct consists of an Fc region of a human antibody with two copies of a peptide sequence linked to each of its IgG1 heavy chains. The peptide sequence was selected to stimulate the thrombopoietin receptor, yet the sequence has no similarity to its endogenous analogue thrombopoietin121. The Fc portion extends the half-life of AMG 531 in the circulation, and the lack of sequence homology to thrombopoietin will ideally prevent the development of crossreactive anti-thrombopoietin antibodies — a serious adverse effect seen with a PEGylated version of thrombopoietin122,123.

A third issue is that for a protein to be physiologically active, post-translational modifications such as glycosylation, phosphorylation and proteolytic cleavage are often required124. These requirements may dictate the use of specific cell types that are capable of expressing and modifying the protein appropriately. In addition, recombinant proteins must be synthesized in a genetically engineered cell type for large-scale production. The host system must produce not only biologically active protein but also a sufficient quantity of this protein to meet clinical demand124. Also, the system must allow purification and storage of the protein in a therapeutically active form for extended periods of time. The protein's stability, folding, and tendency to aggregate may be different in large-scale production and storage systems than in those used to produce the protein for animal testing and clinical trials125,126. Some have proposed engineering host systems that co-express a chaperone or foldase with the therapeutic protein of interest, but these approaches have had limited success.

Potential solutions could include the development of systems in which entire cascades of genes involved in protein folding are induced together with the therapeutic protein; the impetus for this work is the observation that plasma cells, which are natural protein production facilities, use such gene cascades to produce large quantities of monoclonal antibody127. Although bacteria and yeast are generally considered easy to culture, certain mammalian cell types can be more difficult and more costly to culture128. Other methods of production — such as genetically engineered animals and plants — could provide a production advantage. Transgenic cows, goats and sheep have been engineered to secrete protein in their milk, and transgenic chickens that lay eggs filled with recombinant protein are anticipated in the future129. Transgenic plants can inexpensively produce vast quantities of protein without waste or bioreactors130, and potatoes can be engineered to express recombinant proteins and thereby make edible vaccines9. Finally, by using fluid-shaking bioreactors, microlitre-sized culture systems might be able to predict the success of large-scale culture systems and thereby provide substantial cost savings by focusing investment on systems that are more likely to succeed131.

A fourth important challenge is the costs involved in developing protein therapies. For example, switching to recombinant methodology from laborious purification of placentally derived protein has allowed the production of sufficient β-glucocerebrosidase to treat Gaucher's disease in many patients. Even so, the cost of the recombinant protein can be greater than US$ 100,000 per patient per year132.

The example of Gaucher's disease also illustrates aspects of a fifth issue associated with protein therapeutics: ethics (although these ethical issues are not exclusive to protein therapeutics). For example, the possibility of efficacious but expensive protein therapeutics for small but severely ill patient populations, such as patients with Gaucher's disease, can present a dilemma with respect to allocation of financial resources of health-care systems132. In addition, the definition of illness or disease could be challenged by protein therapeutics that can 'improve upon' conditions previously viewed as variants of normal. For example, the definition of short stature may begin to change with the possibility of using growth hormone to increase the height of a child133,134,135,136,137.

Conclusion and future directions

Medicine is approaching a new era in which approaches to manage disease are being made at the level of the genetic and protein information that underlies all biology, and protein therapeutics are playing an increasingly important role. Already, recombinant human proteins make up the majority of FDA-approved biotechnology medicines, which include monoclonal antibodies, natural interferons, vaccines, hormones, modified natural enzymes and various cell therapies. The future potential for such therapies is huge, given the thousands of proteins produced by the human body and the many thousands of proteins produced by other organisms.

Furthermore, recombinant proteins not only provide alternative (or the only) treatments for particular diseases, but can also be used in combination with small-molecule drugs to provide additive or synergistic benefit. Treatment of EGFR-positive colon cancer is illustrative of this point: combination therapy with the small-molecule drug irinotecan, which prevents DNA repair by inhibiting DNA topoisomerase, and the recombinant monoclonal antibody cetuximab, which binds to and inhibits the extracellular domain of the EGFR, results in increased survival in patients with colorectal cancer. The therapeutic synergy between irinotecan and cetuximab may be due to the fact that both drugs inhibit the same EGFR signalling pathway, with one drug (cetuximab) inhibiting the initiation of the pathway and the other drug (irinotecan) inhibiting a target downstream in the pathway82,138,139,140.

The early success of recombinant insulin production in the 1970s created an atmosphere of enthusiasm and hope, which was unfortunately followed by an era of disappointment when the vaccine attempts, non-humanized monoclonal antibodies and cancer trials in the 1980s were largely unsuccessful. Despite these setbacks, significant progress has been made recently. As well as the major successes with protein therapeutics described in this article, new production methods are changing the scale, cost and even route of administration of recombinant protein therapeutics. With the large number of protein therapeutics both in current clinical use and in clinical trials for a range of disorders, one can confidently predict that protein therapeutics will have an expanding role in medicine for years to come.