Biomolecular Structure and Modeling: Problem and Application Perspective

Abstract

The experimental progress described in the previous chapter has been accompanied by an increasing desire to relate the complex three-dimensional (3D) shapes of biomolecules to their biological functions and interactions with other molecular systems. Structural biology, computational biology, genomics, proteomics, bioinformatics, chemoinformatics, and others are natural partner disciplines in such endeavors.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

All things come out of the one, and the one out of all things. Change, that is the only thing in the world which is unchanging.

Heraclitus of Ephesus (550–475 BC).

2.1 Computational Challenges in Structure and Function

2.1.1 Analysis of the Amassing Biological Databases

The experimental progress described in the previous chapter has been accompanied by an increasing desire to relate the complex three-dimensional (3D) shapes of biomolecules to their biological functions and interactions with other molecular systems. Structural biology, computational biology, genomics, proteomics, bioinformatics, chemoinformatics, and others are natural partner disciplines in such endeavors.

Structural biology focuses on obtaining a detailed structural resolution of macromolecules and relating structure to biological function.

Computational biology was first associated with the discipline of finding similarities among nucleotide strings in known genetic sequences, and relating these relationships to evolutionary commonalities; the term has grown, however, to encompass virtually all computational enterprises to molecular biology problems [747].

Comparative genomics — the search and comparison of sequences among species — is a natural outgrowth of the sequencing projects [672]. So are structural and functional genomics, the characterization of the 3D structure and biological function of these gene products [149, 167, 212, 635, 867].

In the fall of 2000, the U.S. National Institute of General Medical Sciences (NIGMS) launched a five-year structural genomics initiative (also called PSI for Protein Structure Initiative) by funding seven research groups aiming to solve collectively the 3D structures of 10,000 proteins, each representing a protein family, over the next decade. This goal of assembling a protein fold library required improvements in both structural biology’s technology and methodology, so the goals included development of methodology and technology to enable high-throughput structure determination and subsequent automation of unique protein structures. After five years, it was realized that those ambitious goals were not met, so both the methodology and structure-determination aims were scaled down significantly in the next phase of funding (2005–2010).

However, despite steady progress [214], by 2008 it became apparent to some leaders in the community, both within and outside these initiatives, that the task at large is much more difficult than previously imagined, perhaps even unattainable, because of the infinitely-large size of the fold space; in addition, the very tight state of funding to NIH-supported scientists in the wake of the Iraq war prompted many scientists to re-evaluate funding such large-scale cataloging projects at the expense of individual, hypothesis-driven research labs who desperately needed the funding. See [993], for example, urging termination of PSI, and an opposing view from some involved scientists [87]. Since then, efforts are mostly shifting to new functional/structural annotations, which may be more meaningful to our goal of understanding the function of genome products.

Proteomics is another current buzzword defining the related discipline of protein structure and function (see [373] for an introduction), and even cellomicshas been introduced.¹ Cellomics reflects the expanded interest of gene sequencers in integrated cellular structure and function. The Human Proteomics Project— a collaborative venture to churn out atomic structures using high-throughput and robotics-aided methods based on NMR spectroscopy and X-ray crystallography, rather than sequences — may well be on its way.

New instruments that have revolutionized genomics known as DNA microarrays, biochips, or gene expression chips (introduced in Chapter 1and Box 1.4) allow researchers to determine which genes in the cell are active and to identify gene networks.

The range of genomic sciences also extends [935] to the metabolome, the endeavor to define the complete set of metabolites (low-molecular cellular intermediates) in cells, tissues, and organs. Experimental techniques for performing these integrated studies are continuously being developed. For example, yeast geneticists have developed a clever technique for determining whether two proteins interact and, thereby by inference, participate in related cellular functions [1283]. Such approaches to proteomics provide a powerful way to discover functions of newly identified proteins. DNA chip technology is also thought to hold the future of individualized health care now coined personalized medicine or pharmacogenomics; see Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151 and Box 1.4. Additionally, as mentioned in the first chapter, genomics and its disciples have already led to drug discovery, as in the notable case of a SARS virus inhibitor [329]; see [191] for the impact of systems biology on drug discovery.

It has been said that current developments in these fields are revolutionary rather than evolutionary. This view reflects the clever exploitation of biomolecular databases with computing technology and the many disciplines contributing to the biomolecular sciences (biology, chemistry, physics, mathematics, statistics, and computer science).Bioinformatics is an all-embracing term for many of these exciting enterprises [571, 881] (structural bioinformatics is an important branch); chemoinformatics has also followed (see Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151) [507]. Some genome-technology company names are indicative of the flurry of activity and grand expectations from our genomic era. Consider titles like Genetics Computer Group, Genetix Ltd., Genset, Protana, Protein Pathways, Inc., Pyrosequencing AB, Sigma-Genosys, or Transgenomic Incorporated. With many companies now in the business of personal genetics (like Navigenics, 23andMe, Knome), even our approach to health, disease prevention, and treatment may be changing.

This excitement in the field’s developments and possibilities is echoed by the chief executive of the software giant Oracle Corp., Lawrence Ellison, who surrounds himself by molecular biologists — the scientists, board members, and fellows of his Ellison Medical Foundation; explaining to a Wall Street Journal reporter his preference of molecular biology over racing sailboats, Ellison said: “The race is more interesting, the people in the race are more interesting and the prize is bigger.” (Wall Street Journal, January 9, 2003). This means a lot from the owner of a multi-million-dollar 90-foot wonder-yacht!

When a new “game”, named Foldit, developed by researchers at the University of Washington, based on the Rosetta @ home software, was introduced to the general public, a ScienceDaily report featured the headline: “Computer Game’s High Score Could Earn The Nobel Prize in Medicine” (May 9, 2008). Whether the serious business of protein folding can be turned into a competitive sport remains to be seen, but the software has surely caught the attention of gamers at large.

Although the number of sequence databases has grown very rapidly and exceeds the amount of structural information, the 1990s saw an exponential rise of structural databases as well.From only 50 solved 3D structures in the Protein Data Bank (PDB) in 1975, the number rose to 500 in 1988; another order of magnitude was reached in 1996 (around 5000 entries), and 50,000 entries were reported before the end of 2008. In fact, the rate of growth of structural information is approaching the rate of increase of amino acid sequence information (see Figure 2.1 and Table 2.1). It is no longer a rare event to see a new crystal structure on the cover of Nature or Science. Now, on a weekly basis, groups compete to have their newly-solved structure on the cover of prominent journals. The number of NMR-deduced structures deposited in the PDB has risen slowly, reflecting roughly 13% of the total solved structures by the end of 2009. (For updated information, check the holdings on RCSB).

Fig. 2.1

The growth of the protein sequence database, NRPR, versus structural database of macromolecules (PDB). See Table 2.1 and www.dna.affrc.go.jp/growth/P-history.html. The NRPR database represents merged, non redundant protein database entries from several databases: PIR, SWISS-PROT, GenPept, and PDB.

Table 2.1 Growth of protein sequence databases.

This trend, coupled with tremendous advances in genome sequencing projects [695], argues strongly for increased usage of computational tools for analysis of sequence/structure/function relationships and for structure prediction and design applications. Thus, besides genomics-based analyses and comparisons, accurate, reliable, and rapid theoretical tools for describing structural and functional aspects of gene products are important. (See [635], for example, for mathematical and computational challenges in genomics).

2.1.2 Computing Structure From Sequence

One of the most successful approaches to date on structure prediction comes from homology modeling (also called comparative modeling) [16, 79].

In general, a large degree of sequence similarity often suggests similarity in 3D structure. It has been reported, for example, that a sequence identity of greater than 40% usually implies more than 90% 3D-structure overlap (defined as percentage of C^α atoms of the proteins that are within 3.5 Å of each other in a rigid-body alignment; see definitions in Chapter 3) [1087]. Thus, sequence similarity of at least 50% suggests that the associated structures are similar overall. Conversely, low sequence similarity generally implies structural diversity. This argument of the poor performance of homology modeling when sequence similarity is < 50% has been used against large-scale initiatives like the PSI [993].

There are many exceptions to these homology/structure similarity relationships, however, as demonstrated humorously in a contest presented to the protein folding community (see Box 2.1).The myoglobin and hemoglobin pair is a classic example where large structural, as well as evolutionary, similarity occurs despite little sequence similarity (20%). Other exceptional examples and various sequence/structure relationships are discussed separately in Chapter 3, as well as Homework 6; see also [436] for examples.

More general than prediction by sequence similarity is structure prediction de novo [79], a Grand Challenge of the field, as described next.

2.2 Protein Folding – An Enigma

2.2.1 ‘Old’ and ‘New’ Views

There has been much progress on the protein folding challenge since Cyrus Levinthal first posed the well-known “paradox” named after him; see [395, 564] for historical perspectives. Levinthal suggested that well-defined folding pathways might exist [743] since real proteins cannot fold by exhaustively traversing through their entire possible conformational space [744]. (See [318, 1294, 1295] for a related discussion on whether the number of protein conformers depends exponentially or non-exponentially on chain length). Levinthal’s paradox led to the development of two views of folding — the ‘old’ and the ‘new’ — which have since merged [313, 314, 362, 707].

The former accents the existence of a specific folding pathway characterized by well-defined intermediates.The latter emphasizes the rugged, heterogeneous multidimensional energy landscape governing protein folding, with many competing folding pathways [1385]. Yet, the boundary between the two views is pliant and the intersection substantial [362, 395, 564]. This integration has resulted from a variety of information sources: theories on funnel-shaped energy landscape (e.g., [179, 314, 949]); folding and unfolding simulations of simplified models (e.g., [396, 430, 509, 660, 749, 861, 1170]), at high temperatures or low pH concentrations (e.g., [707, 1430]); NMR spectroscopic experiments that monitor protein folding intermediates (e.g., [345, 948]); predictions of secondary and/or tertiary structure on the basis of evolutionary information [1086]; and statistical mechanical theories.

Such studies suggest that while wide variations in folding pathways may occur, there exists in general a unifying pattern for the evolution of native-structure contacts, which are encoded in the amino acid sequence of the protein [362]. Thus, while independent pathways can result from occasional misfolding errors that can block certain pathway points and affect intermediates, protein folding is generally an ordered process based on native-like foldon units — cooperative structural units of the native protein — and intermediates. Protein folding landscapes and the different views can therefore be reconciled and interpreted in terms of the combined factors of cooperativity of these structural units, their stepwise stabilization, and the chance occurrence of folding errors.

Interestingly, a recent experimental work focusing on protein folding and unfolding kinetics [1362] confirmed long-standing theoretical hypotheses that protein landscape roughness causes slow folding. Wensley et al. probed the reasons for different folding profiles of the protein domains of α-spectrin, a protein of the intracellular matrix of red blood cells important for membrane elasticity. Prior experimental studies have shown that the R15 domain folds very quickly, roughly 3000 times faster than its homologues R16 and R17. Because the structures are similar, the reason for this behavior was not apparent. By swapping domains through chimeric constructs, the researchers demonstrated that protein landscape roughness, or internal friction resulting from residue-specific interactions that can lead to misdocking of helices, causes slow folding and unfolding of the R16 and R17 domains of α-spectrin. For this membrane protein, slow unfolding kinetics are advantageous because they imply fewer rearrangements during the cell’s lifetime; this, in turn, decreases potential degradation. Thus, besides producing important experimental evidence and insights into folding/unfolding pathways, this work underscores the value of theory in understanding biomolecular structures, dynamics, and pathways. Biomolecular modeling is well on its way to becoming a full partner to experiment and a field on its own right [1464].

2.2.2 Folding Challenges

The great progress in the field can also be seen by evaluations of the highly successful biannual prediction exercises (termed CASP for Critical Assessment of Techniques for Protein Structure Prediction) and associated meetings conducted since 1994. See predictioncenter.org/ for the latest meeting developments, including detailed Proceedings, such as [683, 880, 1270]. Though these events have become high profile endeavors for researchers in the field because success in CASP leads to great recognition, the scientific lessons learned year to year and over time have been invaluable to the protein community. From participation of 35 groups in CASP1 (1994), the number has increased to several hundred.

The goals of CASP are to assess capabilities and limitations of current protein structure prediction, highlight promising areas, pinpoint specific difficulties, and thereby stimulate progress in the field. Specifically, the CASP organizers assign certain proteins for theoretical prediction that protein crystallographers and NMR spectroscopists expect to complete by the next CASP meeting. Prediction assessors then consider several categories of structural prediction tools, for example: template based modeling for tertiary structure prediction; template free modeling for tertiary structure prediction; side chain, loop, and active-site prediction for high resolution models; high accuracy modeling; disordered protein-region identification; domain-boundary identification; function prediction; and more. Evaluators assess how well various in silico approaches such as comparative (homology) modeling or ab initio prediction (i.e., using first principles) perform.

The meetings are important not only for motivating progress in protein prediction but also for revealing important trends concerning the strategies that work well and those that may not be as promising. In particular, the meetings demonstrated that comparative modeling approaches can produce reasonably good structural models, with notably more accurate predictions becoming possible, but that it is still difficult to predict the structure of regions that are substantially different from the target. For example, when the quality of the prediction is characterized in terms of C^α root-mean-square (RMS) deviations, the best values obtained — in the lower part of the range of 2–6 Å — are from the best comparative modeling approaches.

It has also become evident that by combining information from two or more templates and by following homology modeling by clever all-atom refinement, prediction accuracy and quality can be enhanced; all-atom refinement, in particular, has been a stumbling block for a long time, so it is gratifying to finally see progress in this area. Furthermore, the accuracy of models predicted by automatic servers is approaching that of manual manipulation, lending promise to the notion that ultimately every interested individual might be able to automate such protein folding predictions on her/his desktop. Automation and rapid folding simulations can make possible applications to enzyme design, such as done with the Rosetta program; see [609], for example. Still, recognizing entirely novel folds remains a challenge, as well as predicting secondary structures and long-range contacts. Template-free or ab initio modeling has shown more modest general improvements, pointing to needed new ideas; nonetheless it is encouraging that such methods can occasionally perform very well on some targets.

Modeling work in the field is invaluable because it teaches us to ask, and seek answers to, systematic questions about sequence/structure/function relationships and about the underlying forces that stabilize biomolecular structures, especially when using ab initio methods. Still, given the rapid improvements in the experimental arena, the pace at which modeling predictions improve must be expeditious to make a significant contribution to protein structure prediction from sequence.

2.2.3 Folding by Dynamics Simulations?

While molecular dynamics simulations are beginning to approach the timescales necessary to fold small peptides [285] or small proteins [338], we are far from finding the Holy Grail, if there is one [121]. Indeed, ambitious goals declared in the late 1990s like IBM’s desire to fold proteins with the ‘Blue Gene’ petaflop computer (i.e., capable of 10¹⁵ floating-point operations per second) depends on the computational models guiding this ubiquitous cellular process. One of the major computational initiatives to come in its wake is ‘Blue Waters’ by the University of Illinois, NCSA, IBM, and their partners who will launch a petascale computing system in 2011 (see ncsa.uiuc.edu). This effort will likely bring unprecedented computing power that could be exploited to tackle numerous computation-intensive applications involving large and complex systems like biomolecular dynamics; planetary, star and galaxy motion; economical modeling; cyberinfrastructure networks; and more.

For protein folding applications, computational power alone may hardly be sufficient; the well recognized force field approximation remains an issue, as does the need to account for all key factors that dictate folding in vivo. For example, some proteins require active escorts to assist in their folding in vivo. These chaperone molecules assist in the folding and rescue misbehaved polymers. Though many details are not known about the mechanisms of chaperone assistance (see below), we recognize that chaperones help by guiding structure assembly and preventing aggregation of misfolded proteins. For an overview of chaperones, see [519, 762, 1326], for example.

Box 2.1: Paracelsus Challenge

In 1994, George Rose and Trevor Creamer posed a challenge, named after a 16th-century alchemist: change the sequence of a protein by 50% or less to create an entirely different 3D global folding pattern [1066]. Though this challenge might sound not particularly difficult, imagine altering at most 50% of the ingredients for a chocolate cake recipe so as to produce bouillabaisse instead! Rose and Creamer offered a reward of $1000 to entice entrants.

The transmutation was accomplished four years later by Lynne Regan and coworkers [281], who converted the four-stranded β-sheet B1 domain of protein G — which has the β sheets packed against a single helix — into a four-helix bundle of two associating helices called Janus (see Figure 2.2). These contestants achieved this wizardry by replacing residues in a β-sheet-encoding domain (i.e., those with high β-sheet-forming propensities) with those corresponding to the four-helix-bundle protein Rop (repressor of primer). Other modifications were guided by features necessary for Rop stability (i.e., internal salt bridge), and the combined design was guided by energy minimization and secondary-structure prediction algorithms.

Fig. 2.2

Ribbon representations of the B1 domain of IgG-binding protein G and the Rop monomer (first 56 residues), which Janus resembles [281], with corresponding sequences. Half of the protein G β domain (B1) residues were changed to produce Janus in response to the Paracelsus challenge (see Box 2.1 and [1066]). The origin of the residues is indicated by the following color schemes: residues from B1: red; residues from Rop: blue; residues in both: green; residues in neither: black. While experimental coordinates of protein G and Rop are known, the structure of Janus was deduced by modeling. The single-letter amino acid acronyms are detailed in Table 3, Chapter 3

The challenge proposers, though delighted at the achievement they stimulated, concluded that in the future only t-shirt prizes should be offered rather than cash!

2.2.4 Folding Assistants

Current studies on chaperone-assisted folding, especially of the archetypal chaperone duo, the E. coli bacterial chaperonin GroEL and its cofactor GroES, are providing insights into the process of protein folding [387, 762, 1257] (see Figure 2.3 and Box 2.2). The rescue acts of chaperones depend on the subclass of these escorts and the nature of the protein being aided. Some chaperones can assist a large family of protein substrates, while others are more restrictive (see Box 2.2); detailed structural explanations remain unclear. Many families of chaperones are also known, varying in size from small monomers (e.g., 40 or 70 kDa for DnaJ and DnaK of Hsp70) to large protein assemblies (e.g., 810 kDa for GroEL or 880 kDa for the GroEL/GroES complex).

Fig. 2.3

The bullet-shaped architecture of the GroEL/GroES chaperonin/co-chaperonin complex sequence. Overall assembly and dimensions are shown from a side view (left). The top ring is the GroES ‘cap’, and the other layers are GroEL rings. Sidechains are shown in grey. As seen from the top and bottom views (right), a central channel forms in the interior, conducive to protein folding. The protein is organized as three rings that share a 7-fold rotational axis of symmetry (middle), where GroEL contains 14 identical protein subunits assembled in two heptameric rings, and GroES contains 7 smaller identical subunits in its heptamer ring.

The small assistants bind to short runs of hydrophobic residues² to delay premature folding and prevent aggregation. Larger chaperones are likely needed to prevent aggregation of folded compact intermediates in the cell termed ‘molten globules’, requiring a complex trap-like mechanism involving co-chaperones (see also Box 2.2).

Such protein aggregation can occur due to even minor changes in intracellular physiochemical conditions, such as temperature and pressure. Chaperones can rescue active proteins from forming these disrupting aggregates by isolating, unfolding, and translocating them as needed. Together with the cellular machinery for removing damaged proteins, the work of chaperones maintains the pool of active proteins critical to an organism’s life. Misfolded proteins can be the root cause of many debilitating human disorders like Alzheimer’s Disease and Cystic Fibrosis (see separate section). Studies of misfolding are helping to investigate these complex phenomena (e.g., [637]).

Box 2.2: Studies on Protein Escorts

The archetypal chaperone GroEL is a member of a chaperone class termed chaperonins; hsp60 of mitochondria and chloroplasts is another member of this class. These chaperones bind to partially-folded peptide chains and assist in the folding with the consumption of ATP. The solved crystal structure of GroEL/GroES [1405] suggests beautifully, in broad terms, how the large central channel inside a barrel-shaped chaperone might guide protein compaction in its container and monitor incorrect folding by diminishing aggregation (see Figure 2.3).The two-ringed GroEL chaperonin (middle and bottom levels in Figure 2.3) attaches to its partner chaperonin GroES (top ring) upon ATP binding, causing a major conformational change; the size of GroEL nearly doubles, and it assumes a cage shape, with GroES capping over it. This capping prevents the diffusion of partially folded compact intermediates termed ‘molten globules’ and offers them another chance at folding correctly.

Experiments that track hydrogen exchange in unfolded rubisco protein by radioactive tritium (a hydrogen isotope) labeling suggest how misfolded proteins fall into this cavity and are released: a mechanical stretching force triggered by ATP binding partially or totally unfolds the misfolded proteins, eventually releasing the captive protein [1180].

These results also support an iterative annealing (or network model) for chaperone-guided folding, in which the process of forceful unfolding of misfolded molecules, their trapping in the cavity, and their subsequent release is iterated upon until proper folding.

The identification of preferential substrates for GroEL in vivo [570], namely multidomain E. coli proteins with complex αβ folds, further explains the high-affinity interactions formed between the misfolded or partially folded proteins and binding domains of GroEL. These proteins require the assistance of a chaperone because assembly of β-sheet domains requires long-range coordination of specific contacts (not the case for formation of α-helices). Natural substrates for other chaperones, like Eukaryotic type II chaperonin CCT, also appear selective, favoring assistance to proteins like actin [782].

However, such insights into folding kinetics are only the tip of the iceberg. Chaperone types and mechanisms vary greatly, and the effects of macromolecular crowding (not modeled by in vitro experiments) complicate interpretations of folding mechanisms in vivo. Unlike chaperonins, members of another class of chaperones that includes the heat-shock protein Hsp70 bind to exposed hydrophobic regions of newly-synthesized proteins and short linear peptides, reducing the likelihood of aggregation or denaturation. These are classified as ‘stress proteins’ since their amount increases as environmental stresses increase (e.g., elevated temperatures) [762]. Other chaperones are known to assist in protein translocation across membranes.

2.2.5 Unstructured Proteins

Though our discussion has focused on the concept of native folds, not all proteins are intrinsically structured [346]. The intrinsic lack of structure can be advantageous, for example in binding versatility to different targets or in ability to adapt different conformations. Unfolded or non-globular structures are recognized in connection with regulatory functions, such as binding of protein domains to specific cellular targets [346, 1391]. Examples include DNA and RNA-binding regions of certain protein complexes (e.g., basic region of leucine zipper protein GCN4, DNA-binding domain of NFATC1, RNA recognition regions of the HIV-1 Rev protein). Here, the unstructured regions become organized only upon binding to the DNA or RNA target. This folding flexibility offers an evolutionary advantage, which might be more fully appreciated in the future, as more gene sequences that code for unstructured proteins are discovered and analyzed.

2.3 Protein Misfolding – A Conundrum

2.3.1 Prions and Mad Cows

Further clues into the protein folding enigma are also emerging from another puzzling discovery involving certain proteins termed prions. These misfolded proteins — triggered by a conformational change rather than a sequence mutation — appear to be the source of infectious agent in fatal neurodegenerative diseases like bovine spongiform encephalopathy (BSE) or ‘mad cow disease’ (identified in the mid 1980s in Britain), and the human equivalent Creutzfeld-Jacob disease (CJD).³ The precise mechanism of protein-misfolding induced diseases is not known, but connections to neurodegenerative diseases, which include Alzheimer’s, are growing and stimulating much interest in protein misfolding [251, 325, 326, 791].

Stanley Prusiner, a neurology professor at the University of California at San Francisco, coined the term prion to emphasize the infectious source as the protein (‘proteinaceous’), apparently in contradiction to the general notion that nucleic acids must be transferred to reproduce infectious agents. Prusiner won the 1998 Nobel Prize in Physiology or Medicine for this “pioneering discovery of an entirely new genre of disease-causing agents and the elucidation of the underlying principles of their mode of action”.

Prions add a new symmetry to the traditional roles long delegated to nucleic acids and proteins! Since the finding in the 1980s that nucleic acids (catalytic RNAs) can catalyze reactions — a function traditionally attributed to proteins only — the possibility that certain proteins, prions, carry genetic instructions — a role traditionally attributed to nucleic acids — completes the duality of functions to both classes of macromolecules.

2.3.2 Infectious Protein?

Is it possible for an ailment to be transmitted by ‘infectious proteins’ rather than viruses or other traditional infectious agents? The prion interpretation for the infection mechanism remains controversial for lack of clear molecular explanation. In fact, one editorial article stated that “whenever prions are involved, more open questions than answers are available” [9]. Yet the theory is winning more converts with laboratory evidence that an infectious protein that causes mad cow disease also causes a CJD variant in mice [1151]. These results are somewhat frightening because they suggest that the spread of this illness from one species to another is easier than has been observed for other diseases.

The proteinaceous theory suggests that the prion protein (see Figure 2.4) in the most studied neurodegenerative prion affliction, scrapie (long known in sheep and goats), becomes a pathologic agent upon conversion of one or more of its α-helical regions into β-regions (e.g., parallel β-helix [1371]); once this conformational change occurs, the conversion of other cellular neighbors proceeds by a domino-like mechanism, resulting in many abnormally-folded molecules which eventually reap havoc in the mammal. This protein-only hypothesis was first formulated by J.S. Griffith in 1967, but Prusiner first purified the hypothetical abnormalprotein thought to cause BSE. New clues are rapidly being added to this intriguing phenomenon (see Box 2.3).

Fig. 2.4

Structure of the prion protein.

Both the BSE and CJD anomalies implicated with prions have been linked to unusual deposits of protein aggregates in the brain. (Recent studies on mice also open the possibility that aberrant proteins might also accumulate in muscle tissue).It is believed that a variant of CJD has caused the death of dozens of people in Britain (and a handful in other parts of the world) since 1995 who ate meat infected with BSE, some only teenagers. Recent studies also suggest that deaths from the human form of mad cow disease could be rising significantly and spreading within Europe as well as to other continents.

Since the incubation period of the infection is not known — one victim became a vegetarian 20 years before dying of the disease — scientists worry about the extent of the epidemic in the years to come. The consequences of these deaths have been disastrous to the British beef industry and have led indirectly to other problems (e.g., the 2001 outbreak of foot-and-mouth disease, a highly infectious disease of most farm animals except horses). The panic has not subsided, as uncertainties appear to remain regarding the safety of various beef parts, as well as sheep meat, and the possible spread of the disease to other parts of the world.

2.3.3 Other Possibilities

Many details of this intriguing prion hypothesis and its associated diseases are yet to be discovered and related to normal protein folding. Some scientists believe that a lurking virus or virino (small nonprotein-encoding virus) may be involved in the process, perhaps stimulating the conformational change of the prion protein, but no such evidence has yet been found. Only creation of an infection de novo in the test tube is likely to convince the skeptics, but the highly unusual molecular transformation implicated with prion infection is very difficult to reproduce in the test tube.

Box 2.3: Prion: Structural Evidence

The detailed structural picture associated with the prion conformational change is only beginning to emerge as new data appear [10]. In 1997, Kurt Wütrich and colleagues at the Swiss Federal Institute of Technology in Zurich reported the first NMR solution structure of the 208-amino acid glycoprotein “prion protein cellular” PrP^C anchored to nerve cell membranes. The structure reveals a flexibly disordered assembly of helices and sheets (see Fig. 2.4). This organization of the harmless protein might help explain the conversion process to its evil isoform PrP^Sc. It has been suggested that chaperone molecules may bind to PrP^C and drive its conversion to PrP^Sc and that certain membrane proteins may also be involved in the transformation.

In early 1998, a team from the University of California at San Francisco discovered a type of prion, different from that associated with mad cow disease, that attaches to a major structure in neuron cells and causes cells to die by transmitting an abnormal signal. This behavior was observed in laboratory rats who quickly died when a mutated type of prion was placed into the brains of newborn animals; their brains revealed the abnormal prions stuck within an internal membrane of neuron cells. The researchers believe that this mechanism is the heart of some prion diseases. They have also found such abnormal prions in the brain tissue of patients who died from a rare brain disorder called Gerstmann-Straussler-Scheinker disease (GSS) — similar to Creutzfeld-Jacob disease (CJD) — that destroys the brain.

Important clues to the structural conversion process associated with prion diseases were further offered in 1999, when a related team at UCSF, reported the NMR structure of the core segment of a prion protein rPrP that is associated with the scrapie prion protein PrP^Sc [602, 777]. The researchers found that part of the prion protein exhibits multiple conformations. Specifically, an intramolecular hydrogen bond linking crucial parts of the protein can be disrupted by a single amino acid mutation, leading to different conformations. This compelling evidence on how the molecule is changed to become infectious might suggest how to produce scrapie-resistant or BSE-resistant species by animal cloning.

Prion views from several organisms (including human and cow) have been obtained [1429], allowing analyses of species variations, folding, and misfolding relationships; see [1371], for example. This high degree of similarity across species is shown in Figure 2.4.

Still, until prions are demonstrated to be infectious in vivo, the proteinaceous hypothesis warrants reservation. Clues into how prions work may emerge from parallel work on yeast prions, which unlike their mammalian counterparts do not kill the organism but produce transmitted heritable changes in phenotype; many biochemical and engineering studies are underway to explore the underlying mechanism of prion inheritance.

2.3.4 Other Misfolding Processes

There are other examples of protein misfolding diseases (e.g., references cited in [325, 326, 505, 791]). The family of amyloid diseases includes Alzheimer’s, Parkinson’s, and type II (late-onset) diabetes. For example, familial amyloid polyneuropathy is a heritable condition caused by the misfolding of the protein transthyretin. The amyloid deposits that result interfere with normal nerve and muscle function.

Dobson [325] intriguingly suggests that understanding the evolution of proteins holds the key to protein misfolding diseases. Namely, he argues that since evolutionary processes have selected sequences of amino acids that form close-packed, globular proteins, the effectively irreversible formation of amyloid fibrils reflects a conversion of proteins to their ‘primordial’ rather than evolved states, possibly from aging-induced mutations that destabilize native proteins.

Indeed, many protein misfolding diseases are strongly associated with aging, suggesting that the cell’s ability to monitor misfolding and prevent aggregation deteriorate with age. Fortunately, recent biophysical and computational techniques are leading to an increased understanding of what triggers protein misfolding and what the intrinsic and extrinsic factors that contribute to the process in vivo, though we are far from rational design of therapeutic intervention [791]. Computational models of protein misfolding, in particular, can help relate systematically changes in temperature-dependent pathways and aggregation to observed phenomena.

As in mad cow disease, a molecular understanding of the misfolding process may lead to treatments of the disorders. In the case of familial amyloid polyneuropathy, research has shown that incorporating certain mutant monomers in the tetramer protein transthyretin reduces considerably the formation of amyloid deposits (amyloid fibrils); moreover, incorporating additional mutant monomers can prevent misfolding entirely [505]. These findings suggest potential therapeutic strategies for amyloid and related misfolding disorders. See also [983] for a pharmacological approach for treating human amyloid diseases by using a small-molecule drug that targets a protein present in amyloid deposits; the drug links two pentamers of that protein and leads to its rapid clearance by the liver.

Studies also suggest that misfolded proteins generated in the pathway of protein folding can be dangerous to the cell and cause harm (whether or not they convert normal chains into misfolded structures, as in prion diseases) [183, 1325]. The cellular mechanisms associated with such misfolded forms and aggregates are actively being pursued, including by modeling [637].

2.3.5 Deducing Function From Structure

Having the sequence and also the 3D structure at atomic resolution, while extremely valuable, is only the beginning of understanding biological function. How does a complex biomolecule accommodate its varied functions and interactions with other molecular systems? How sensitive is the 3D architecture of a biopolymer to its constituents?

Despite the fact that in many situations protein structures are remarkably stable to tinkering (mutations), their functional properties can be quite fragile. In other words, while a protein often finds ways to accommodate substitutions of a few amino acids so as not to form an entirely different overall folding motif [205], even the most minute sequence changes can alter biological activity significantly. Mutations can also influence the kinetics of the folding pathway.

An example of functional sensitivity to sequence is the altered transcriptional activity of various protein/DNA complexes that involve single base changes in the TATA-box recognition element and/or single protein mutations in TBP (TATA-Box binding protein) [971]. For example, changing just a single residue in the common nucleotide sequence of TATA-box element, TATAAAAG, to TAAAAAAG impairs binding to TBP and hence disables transcriptional activity.

In principle, theoretical approaches should be able to explain these relations between sequence and structure from elementary physical laws and knowledge of basic chemical interactions. In practice, we are encountering immense difficulty pinpointing what Nature does so well. After all, the notorious “protein folding” problem is a challenge to us, not to Nature.

Much work continues on this active front.

2.4 From Basic to Applied Research

An introductory chapter on biomolecular structure and modeling is aptly concluded with a description of the many important practical applications of the field, from food chemistry to material science to drug design. A historical perspective on drug design is given in Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151. Here, we focus on the current status of drug development as well as other applied research areas that depend strongly on progress in molecular modeling. Namely, as biological structures and functions are being resolved, natural disease targets that affect the course of disease can be proposed. Such new treatments can be approached both from the traditional drug design model which seeks inhibitors to specific targets (e.g., reviewed in [453, 913, 1193, 1447]) or from a systems biology approach which attempts to modify response of genes, proteins, and metabolites by integrating organ and system-level modeling [191, 278, 649]. Other biological and polymer targets, such as the ripening genes of vegetables and fruit or strong materials, can also be manipulated to yield benefits to health, technology, and industry.

2.4.1 Rational Drug Design: Overview

The concept of systematic drug design, rather than synthesis of compounds that mimic certain desired properties, is only about 50 years old (see Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151). Gertrude Elionand George Hitchings of Burroughs Wellcome, who won the 1988 Nobel Prize in Physiology or Medicine, pioneered the field by creating analogues of the natural DNA bases in an attempt to disrupt normal DNA synthesis. Their strategies eventually led to a series of drugs based on modified nucleic-acid bases targeted to cancer cells. Today, huge compound libraries are available for systematic screening by various combinatorial techniques, robotics, other automated technologies, and various modeling and simulation protocols (see Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151).

Rational pharmaceutical design has now become a lucrative enterprise. The sales volume for the world’s best seller prescription drug in 1999, Prilosec (for ulcer and heartburn), exceeded six billion dollars. A vivid description of the climate in the pharmaceutical industry and on Wall Street can be found in The Billion-Dollar Molecule: One Company’s Quest for the Perfect Drug [1363]. This thriller describes the racy story of a new biotech firm for drugs to suppress the immune system, specifically the discovery of an alternative treatment to Cyclosporin, medication given to transplant patients. Since many patients cannot tolerate cyclosporin, an alternative drug is often needed.

Tremendous successes in 1998, like Pfizer’s anti-impotence drug Viagra and Entre-Med’s drugs that reportedly eradicated tumors in mice, have generated much excitement and driven sales and earnings growth for drug producers. A glance at the names of biotechnology firms is an amusing indicator of the hope and prospects of drug research: Biogen, Cor Therapeutics, Genetech, Genzyme, Immunex, Interneuron Pharmaceuticals, Liposome Co., Millennium Pharmaceuticals, Myriad Genetics, NeXstar Pharmaceuticals, Regeneron Pharmaceuticals, to name a few. Other success stories involve a small-molecule inhibitor of the SARS virus [329], glutamate nanosensors to monitor neurologic functions whose malfunction can lead to neurodegenerative disorders [933], and agonists to treat anxiety and depression [111]. Yet, both the monetary cost and development time required for each successful drug remains very high [39, 160], and great successes are now few and far between; see end of chapter for further discussion.

2.4.2 A Classic Success Story: AIDS Therapy

2.4.2.1 HIV Enzymes

A spectacular example of drugs made famous through molecular modeling successes are inhibitors of the two viral enzymes HIV protease (HIV: human immunodeficiency virus) andreverse transcriptase for treating AIDS, acquired immune deficiency syndrome.

First hints of AIDS were reported in the summer of 1981, in clusters of gay men in large American cities; these groups exhibited severe symptoms of infection by certain pneumonia combined with those from Kaposi’s sarcoma (KS) cancer. Now considered among the most catastrophic pandemics to strike humankind, this infectious disease is caused by an insidious retrovirus. (See perspectives on the evolution of this pandemic, including treatment and prevention in [253, 611, 1221] and a personal reflection by Robert Gallo who was instrumental in identifying the retrovirus culprit [434]). Such a virus can convert its RNA genome into DNA, incorporate this DNA into the host cell genome, and then spread from cell to cell. To invade the host, the viral membrane of HIV must attach and fuse with the victim’s cell membrane; once entered, the viral enzymes reverse transcriptase and integrase transform HIV’s RNA into DNA and integrate the DNA into that of the host [529].

Current drugs inhibit enzymes that are key to the life cycle of the AIDS virus (see Figure 2.5).Protease inhibitors like Indinavir, Saquinavir, Ritonavir, Nelfinavir, etc. block the activity of proteases, protein-cutting enzymes that help a virus mature, reproduce, and become infectious [227]. Reverse transcriptase (RT) inhibitors block the action of an enzyme required by HIV to make DNA from its RNA [1045].However, eradication of the disease by a preventive HIV vaccine has so far been largely unsuccessful due to the complex biology and life cycle of the virus [91, 253, 375, 611, 612]. Still, existing medications can give AIDS patients a life.

Fig. 2.5

Examples of AIDS drug targets — the HIV protease inhibitor and reverse transcriptase (RT) — with corresponding designed drugs. The protease inhibitor Indinavir (crixivan) binds tightly to a critical area of the dimer protease enzyme (HIV-2, 198 residues total shown here [227]), near the flaps (residues 40 to 60 of each monomer), inducing a conformational change (flap closing) that hinders enzyme replication; intimate interactions between the ligand and enzyme are observed in residues 25 and 50 in each protease monomer. The non-nucleoside RT inhibitor 1051U91 (a nevirapine analogue), approved for use in combination with nucleoside analogue anti-HIV drugs like AZT, binds to a location near the active site of RT that does not directly compete with the oligonucleotide substrate. The large RT protein of 1000 residues contains two subdomains (A and B).

2.4.2.2 AIDS Drug Development

One of the most commonly used drug cocktails is the triplet drug combination of a protease inhibitor like indinavir with the two nucleoside analogues like AZT (Zidovudine, or 3′-azido-3′-deoxythymidine) and 3TC. Another commonly prescribed regimen utilizes two nucleoside analogues and one non-nucleoside RT inhibitors (see below). More than one drug is needed because mutations in the HIV enzymes can confer drug resistance; thus, acting on different sites as well as on different HIV proteins increases effectiveness of the therapy.

The two types of RT blocker mentioned above are nucleoside analogues and non-nucleoside inhibitors. Members of the former group (Zidovudine or AZT, Didanosine, Zalcitabine, Stavudine, etc.) interfere with the HIV activity by replacing a building block used to make DNA from the HIV RNA virus with an inactive analog and thereby prevent accurate decoding of the viral RNA. Non-nucleoside RT inhibitors (e.g., Nevirapine, Delavirdine, and Efavirenz) are designed to bind with high affinity to the active site of reverse transcriptase and therefore physically interfere with the enzyme’s action.

Design of such drugs was made possible in part by molecular modeling due to the structure determination of the HIV protease by X-ray crystallography in 1989 and RT a few years later [1218]. Figure 2.5 shows molecular views of these HIV enzymes complexed with drugs.

Besides the HIV protease and reverse transcriptase, a third target is the HIV integrase, which catalyzes the integration of a DNA copy of the viral genome into the host cell chromosomes. Scientists at Merck identified several years ago 1,3-diketo acid integrase inhibitors that block strand transfer, one of the two specific catalytic functions of HIV-1 integrase [536]; this function has not been affected by previous inhibitors. This finding paved the way for developing effective integrase inhibitors: Raltegaravir was approved in this class in 2007.

2.4.2.3 AIDS Drug Limitations

Much progress has been made in this area since the first report of the rational design of such inhibitors in 1990 [1054] (see [253, 434, 611, 612, 1221] for reviews). In fact, the dramatic decline of AIDS-related deaths by such drug cocktails can be attributed in large part to these new generation of designer drugs (see Box 2.4) since the first introduction of protease inhibitors in 1996. Indeed, the available triplet drug cocktails, of protease inhibitors and nucleoside analogues RT inhibitors, or nucleoside analogues and non-nucleoside RT inhibitors, have been shown to virtually suppress HIV, making AIDS a manageable disease.

However, the cocktails are not a cure. The virus returns once patients stop the treatment, and the enormous genetic diversity of mutations that occur enable HIV to reduce the effectiveness of treatment. Indeed, in very heavily treated patients, as many as one quarter of the amino acids (25 out of 99) of the viral protein HIV-1 protease can be mutated, but the enzyme continues to function. Moreover, the window of opportunity for the immune system to clear the initial infection is very narrow, because the virus quickly integrates itself into the host. The mechanisms of drug resistant mutations and the interactions among them are still not well understood despite enormous amount of research, and fundamental questions about the progression of HIV disease and the host response to the virus remain unanswered [91, 612].

In addition, few countries in the developing world, like Africa, can afford the virus suppressing drugs; the drug-cocktail regimen is complex, requiring many daily pills taken at multiple times and separated from eating, most likely for life; serious side effects also occur. For example, we now know that nucleoside analogues inhibit a variety of DNA polymerization reactions, in addition to those of the HIV-1 RT, and are thus associated with serious side effects.

In certain parts of the world, the situation is profoundly distressing: the life expectancy of patients living with HIV/AIDS in many African countries has fallen to 40 years of age today, a drastic difference from the age in the pre-AIDS era, and the number continues to drop.⁴ Though in the developed world, AIDS is no longer a death sentence, the incidence of new infections is alarming in certain urban areas. For example, the U.S. Center for Disease Control and Prevention reported in late August 2008 that HIV is spreading in New York City at three times the national rate (72 versus 23 new cases per 100,000 people).

2.4.2.4 Lurking Virus

As mentioned, even available treatments cannot restore the damage to the patient’s immune system; the number of T-cell (white blood cells), which HIV attaches itself to, is still lower than normal (which lowers the body’s defenses against infections), and there remain infected immune cells that the drugs cannot reach because of integration. Thus, new drugs are being sought to interrupt the first step in the viral life cycle — binding to a co-receptor on the cell surface to rid the body of the cell’s latent reservoirs of the HIV virus, to chase the virus out of cells where it hides for subsequent treatment, or to drastically reduce the HIV reservoir so that the natural immune defenses can be effective. New structural and mechanistic targets are currently being explored (see Box 2.4). Some of the newest drugs under development include low-cost microbicidal drugs which can be topically applied prior to sexual contact to prevent, or directly destroy HIV [84].

A better understanding of the immune-system mechanism associated with AIDS, for example, may help explain how to prime the immune system to recognize an invading AIDS virus. Unlike traditional AIDS drug cocktails which inhibit division of already infected cells, fusion (or entry) inhibitors define another class of drugs that seek to prevent HIV from entering the cell membrane. This entry, called fusion, releases the virus’s genetic material and allows it to replicate. The promising drug T-20 or Enfuvirtide (which must be injected into the skin) is a member of fusion-inhibitor or entry-inhibitor drugs that, when added to a combination of standard drugs, can significantly reduce HIV levels in the blood. Another such entry-inhibitor is Maraviroc, a CCR5 antagonist (see Box 2.4).

As manifested by its complex components of invasion that include the fusion apparatus, the AIDS virus has developed a complex, tricky, and multicomponent-protection infection machinery, as well as drug-resistant defense.

Besides integrase and fusion inhibitors, among the newer drugs to fight AIDS being developed are immune stimulators and antisense drugs. The former stimulate the body’s natural immune response, and the latter mimic the HIV genetic code and prevent the virus from functioning.

Box 2.4: Fighting AIDS

AIDS drugs attributed to the success of molecular modeling include AZT (Zidovudine) sold by Bristol-Myers Squibb, and the newer drugs Viracept (Nelfinavir) made by Agouron Pharmaceuticals, Crixivan (Indinavir) by Merck & Company, and Amprenavir discovered at Vertex Pharmaceuticals Inc. and manufactured by Glaxo Wellcome. Amprenavir, in particular, approved by the U.S. Government in April 1999, is thought to cross the ‘blood-brain barrier’ so that it can attack viruses that lurk in the brain, where the virus can hide. This general class of inhibitors has advanced so rapidly that drug-resistant AIDS viruses have been observed.

Structural investigations are probing the structural basis for the resistance mechanisms, which remain mysterious, particularly in the case of nucleoside analogue RT inhibitors like AZT [700]. The solved complex of HIV-1 reverse transcriptase [576] offers intriguing insights into the conformational changes associated with the altered viruses that influence the binding or reactivity of inhibitors like AZT and also suggests how to construct drug analogues that might impede viral resistance.

Basic research on the virus’s process of invading host cells — by latching onto receptors (e.g., the CD4 glycoprotein, which interacts with the viral envelope glycoprotein, gp120, and the transmembrane component glycoprotein, gp41), and co-receptors (e.g., CCR5 and CXCR4) — may also offer treatments, since developments of disease intervention and vaccination are strongly aided by an understanding of the complex entry of HIV into cells; see [687] for example.

The HIV virus uses a spear-like agent on the virus’s protein coat to puncture the membrane of the cells which it invades; vaccines might be designed to shut the chemical mechanism or stimuli that activate this invading harpoon of the surface protein. The solved structure of a subunit of gp41, for example, has been exploited to design peptide inhibitors that disrupt the ability of gp41 to contact the cell membrane [393]. A correlation has been noted, for example, between co-receptor adaptation and disease progression.

Novel techniques for gene therapy for HIV infections are also under development, such as internal antibodies (intrabodies) against the Tat protein, a vehicle for HIV infection of the immune cells; it is hoped that altered T-cells that produce their own anti-Tat intrabody will lengthen the survival time of infected cells or serve as an HIV ‘dead-end’.

Other clues to AIDS treatments may come from the finding that HIV-1 originally came from a subspecies of chimpanzees [440]. Since chimps have likely carried the virus for hundreds of thousands of years but have not become ill from it, understanding this observation might help fight HIV-pathogeny in humans. Help may also come from the interesting finding that a subset of humans have a genetic mutation (32 bases deleted from the 393 of gene CCR5) that creates a deficient T-cell receptor; this mutation intriguingly slows the onset of AIDS. Additionally, a small subset of people is endowed with a huge number of helper (CD4) T-cells which can coordinate an attack on HIV and thus keep the AIDS virus under exquisite control for many years; such people may not even be aware of the infection for years.

2.4.2.5 Vaccine?

Still, many believe that only an AIDS vaccine offers true hope against this deadly disease. Yet the research on vaccines trails behind the development of drugs, which offer much greater financial incentives and lower risks than vaccines. The vaccine AIDSVAX by the California-base company VaxGen, a genetically amplified version of a single protein from the outer shell of the AIDS virus, offered only limited protection.

Another vaccine under development by an Oxford team (part of the International AIDS Vaccine Initiative) is exploiting for vaccine development the immunological data gleaned from Nairobi women who have remained unaffected by AIDS despite many years of high-risk sexual behavior. These women’s T-cells were found to fight off the disease by attacking two particular proteins produced by the AIDS virus. The DNA sequences making those proteins were subsequently identified and used to create a vaccine specific to viral infections in East Africa; besides the DNA component associated with the relevant genes, the vaccine was amplified with a benign virus copy with same DNA sequences inserted.

Early attempts to target the outer protein envelope of HIV, gp120, turned disappointing, likely because not all virus particles were neutralized. Other vaccines have also been developed, but response is far from ideal. Thus, the announcement in September 2009 that, after 20 years of constant failure, a vaccine which blends two experimental vaccines that had previously failed to work on their own — Sanofi-Pasteur’s ALVAC canary pox/HIV vaccine and VaxGen’s AIDSVAX — offered some protection by reducing the rate of infection by 30% generated great excitement. However, results puzzle researchers because, while reducing infection, the combination vaccine does not reduce the virus levels in the blood. Research is ongoing.

In general, vaccine research experience suggests that a constant level of exposure (e.g., booster shots) is needed to yield immunity, and this defeats the main vaccine advantage of convenience and low cost. Observations also suggest that combinations of vaccines may be needed, since the HIV virus mutates and replicates quickly.⁵ Still, it is hoped that therapeutic vaccination in combination with anti-HIV-1 drug treatment, even if it fails to eradicate infection, will suppress AIDS infection and the rate of transmission, and ultimately decrease the number of AIDS deaths substantially. One of the recent vaccine initiatives includes inducing primary T-cell mediated response to decrease the probability of initial infection [91, 612].

Besides focusing on the role of T cells in the control of the HIV disease progression, other current efforts are attempting to understand the complex immune-response behavior by various participants in the vaccination trials and to broaden the field of HIV vaccine research from new perspectives [375]. Very recently, a novel approach using using RNA silencing has shown promise, by suppressing certain host viral genes crucial to the virus’s replication; the small RNA molecules were delivered to the T cells via a small peptide [686].

However, it is becoming apparent that large resources and enormous leaps — in many fields like genetics, cellular and systems biology — are needed to succeed in preventing this devastating disease.

2.4.3 Other Drugs and Future Prospects

2.4.3.1 Success Stories

Another example of drug successes based on molecular modeling is the design of potent thrombin inhibitors. Thrombin is a key enzyme player in blood coagulation, and its repressors are being used to treat a variety of blood coagulation and clotting-related diseases. Merck scientists reported [161] how they built upon crystallographic views of a known thrombin inhibitor to develop a variety of inhibitor analogues. In these analogues, a certain region of the known thrombin inhibitor was substituted by hydrophobic ligands so as to bind better to a certain enzyme pocket that emerged crucial for the fit. Further modeling helped select a subset of these ligands that showed extremely compact thrombin/enzyme structures; this compactness helps oral absorption of the drug. The most potent inhibitor that emerged from these modeling studies has demonstrated good efficacy on animal models [161].

Other examples of drugs developed in large part by computational techniques include the SARS virus inhibitor [329], glutamate nanosensors to monitor neurologic functions [933], agonists to treat anxiety and depression [111], the antibacterial agent Norfloxacin of Kyorin Pharmaceuticals (noroxin is one of its brand names), glaucoma treatment Dorzolamide (“Trusopt”/Merck), Alzheimer’s disease treatment Donepezil (“Aricept”/ Eisai), and migraine medicine Zolmitriatan (“Zomig”) discovered by Wellcome and marketed by Zeneca [160]. The headline-generating drug that combats impotence (Viagra) was also found by a rational drug approach. It was interestingly an accidental finding: the compound had been originally developed as a drug for hypertension and then angina.

There are also notable examples of herbicides and fungicides that were successfully developed by statistical techniques based on linear and nonlinear regression and classical multivariate analysis (or QSAR, see Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151):the herbicide metamitron — bestseller in 1990 in Europe for protecting sugar beet crops — was discovered by Bayer AG in Germany.

2.4.3.2 Impact of Technology and Modeling

With these new discoveries, we are enjoying improved treatments for cancer, AIDS, heart disease, Alzheimer and Parkinson’s disease, migraine, arthritis, and many more ailments. As new drug targets are being identified — such as new potential sites for antibiotics on the ribosome revealed by a combination of crystallography and bioinformatics, and new protein interfaces within the influenza virus’s RNA polymerase that might be targeted to disrupt polymerase assembly and thus viral replication, as revealed by crystallographic views of RNA polymerase — new opportunities for drug design by modeling become available.

In fact, high-throughput technologies that rely on progress in many fields from genomics to proteomics to imaging can now be processed through the new fields of knowledge-based biological information, like bioinformatics [571, 881] and chemoinformatics [507].Improved modeling and library-based techniques, coupled with robotics and high-speed screening, are also likely to increase the demand for faster and larger-memory computers. “In a marriage of biotech and high tech,” wrote the New York Times reporter Andrew Pollack in 1998, “computers are beginning to transform the way drugs are developed, from the earliest stage of drug discovery to the late stage of testing the drugs in people”.

2.4.3.3 Declining Productivity

However, since the above statement was made, progress in drug development has not exhibited the growth hoped for by emerging technologies. In fact, the industry has actually contracted from a peak of around 50 new approved pharmaceutical agents, also known as new molecular entities (NMEs), in 1996 to half that value in 2008 and 2009 [885]. This slump is even more serious considering that Research and Development (R&D) costs have increased dramatically during this period. Thus, the average cost of $500-800 million and time of 12–15 years required to develop a single drug remain extremely high.

There are many reasons for this disappointing trend.

First, due to safety issues discovered after drugs were approved,⁶ the FDA has implemented continuously rising risk-averse requirements for drug approval, and these modified protocols affect all stages of drug development: discovery and preclinical testing, clinical studies, and registration/approval process.

Second, discovery of new drugs may be more difficult since many of the simple targets/strategies were already considered; this is not unlike the search for new protein folds, which has turned out to be more challenging than originally expected. This difficulty is also reflected by the smaller percentage of truly innovative new drugs among the NMEs.

Third, the “patent cliff” is also affecting this reduction in major pharmaceutical R&D productivity. This cliff refers to loss of revenue when patents for blockbuster drugs expire. These expirations are hitting many companies in a relatively short period around 2010.⁷ These patent expirations lead to sharp profit declines if the company’s drug labs are barren, with no blockbuster substitutes coming out of the pipeline by patent expiration time; in turn, these losses reduce investments in new drug development.

Though a handful of new biologics — biomolecules derived from living cells instead of traditional small-molecule drugs (e.g., Enbrel for rheumatoid arthritis, Herceptin for breast cancer) — are being approved and these help make up for the dip in traditional small-molecule drugs, the long-anticipated breakthroughs in drug development due to high-throughput, genomics-based approaches and biotech agents have not yet been realized. See Chapter 15Similarity and Diversity in Chemical Designchapter.15.1151 for examples of biologics and further discussion of the computational challenges in drug design.

Perhaps, as the new director of NIH exclaimed in January 2010, “The power of the molecular approach to health and disease has steadily gained momentum over the past several decades and is now poised to catalyze a revolution in medicine.” [257]. However, it is becoming clear that such revolutionary advances in drug development, anticipated in the next decade from a combination of high-throughput approaches, biologics, pharmacogenomics, and other innovations, require new integrated paradigms to manage the complex scientific, technological, economic, and business factors involved and reverse the ebbing trends. A better yield of innovative and cost-effective pharmaceutical agents might also alleviate the industry’s political challenges, associated with inadequate availability of drugs to the world’s poor population.

2.4.4 Gene Therapy – Better Genes

Looking beyond drugs, gene therapy is another approach that is benefiting from key advances in biomolecular structure/function studies.Gene therapy attempts to compensate for defective or missing genes that give rise to various ailments — like hemophilia, the severe combined immune deficiency SCID,sickle-cell anemia, cystic fibrosis, and Crigler-Najjar (CN) syndrome — by trying to coerce the body to make new, normal genes. This regeneration is attempted by inserting replacement genes into viruses or other vectors and delivering those agents to the DNA of a patient (e.g., intravenously). However, delivery control, biological reliability, as well as possible unwelcome responses by the body against the foreign invader, remain serious technical hurdles.

One of the classic gene therapy strategies involves direct injection of the thymidine kinase (TK) gene vector into tumors of cancer patients to control cell replication. When the TK gene is expressed, cancer cells can be killed after administration of Gancyclovir, which is converted by TK into a toxic nucleotide. This approach was initially used in aggressive brain tumors (glioblastma multiforme) and more recently for locally recurrent prostate, breast, and colon tumors, among others. See Box 2.5 for other examples of gene therapy.

The first death in the fall of 1999 of a gene therapy patient treated with the common fast-acting weakened cold virus adenovirus led to a barrage of negative publicity for gene therapy.⁸ However, the first true success of gene therapy was reported four months later: the lives of most infants who would have died of the severe immune disorder SCID (and until then lived in airtight bubbles to avoid the risk of infection) were not only saved, but able to live normal lives following gene therapy treatments that restore the ability of a gene essential to make T cells [208]. Unfortunately, complications arose in several of the treated infants by late 2002, including deaths from gene therapy and as well as acquired leukemia [624]. (see Box 2.5).

Though such medical advances appear just short of a miracle, it remains to be seen how effective gene therapy will be on a wide variety of diseases and over a long period. Still, by early 2010, gene therapy treatments may have turned the corner. Small successes have accumulated, for treating children with a fatal brain disease (X-linked adrenoleukodystrophy or ADL) by inserting a corrective gene into the blood cells [890]; a rare form of inherited blindness that strikes at infancy (Leber’s congenital amaurosis or LCA), by injecting the eye with a harmless virus carrying a gene coding for an enzyme necessary for making a light-sensing pigment [814]; and the severe immune disease SCID or “Bubble Boy”, by replacement of the enzyme adenosine deaminase [14]. Thus, cautious optimism is certainly warranted. And for the patients who gained site or normal function after living with serious genetic disorders, gene therapy can be short of a miracle.

A related technique for designing better genes is another relatively new approach known as directed molecular evolution. Unlike protein engineering, in which natural proteins are improved by making specific changes to them, directed evolution involves mutating genes in a test tube and screening the resulting (‘fittest’) proteins for enhanced properties. Companies specializing in this new Darwinian mimicking (e.g., Maxigen, Diversa, and Applied Molecular Evolution) are applying such strategies in an attempt to improve the potency or reduce the cost of existing drugs, or improve the stain-removing ability of bacterial enzymes in laundry detergents. Beyond proteins, such ideas might also be extended to evolve better viruses to carry genes into the body for gene therapy or evolve metabolic pathways to use less energy and produce desired nutrients (e.g., carotenoid-producing bacteria).

Box 2.5: Gene Therapy Examples

A prototype disease model for gene therapy is hemophilia, whose sufferers lack key blood-clotting protein factors. Specifically, Factor VIII is missing in hemophilia A patients (the common form of the disease); the much-smaller Factor IX is missing in hemophilia B patients (roughly 20% of hemophiliacs in the United States).

Early signs of success in treatment of hemophilia B using adeno-associated virus (a vector not related to adenovirus, which is slower acting and more suitable for maintenance and prevention) were reported in December 1999. However, introducing the much larger gene needed for Factor VIII, as required by the majority of hemophiliacs, is more challenging. Here, the most successful treatments to date only increase marginally this protein’s level. Yet even those minute amounts are reducing the need for standard hemophilia treatment (injections of Factor IX) in these patients.

Larger vectors to stimulate the patient’s own cells to repair the defective gene are thus sought, such as retroviruses (e.g., lentiviruses, the HIV-containing subclass), or non-virus particles, like chimeraplasts (oligonucleotides containing a DNA/RNA blend), which can in theory correct point mutations by initiating the cell’s DNA mismatch repair machinery.

An interesting current project involving chimeraplasts is being tested in children of Amish and Mennonite communities to treat the debilitating Crigler-Najjar (CN) syndrome. Sufferers of this disease lack a key enzyme which break down the toxic waste product bilirubin, which in the enzyme’s absence accumulates in the body and causes jaundice and overall toxicity. Children with CN must spend up to 18 hours a day under a blue light to clear bilirubin and seldom reach adulthood, unless they are fortunate to receive and respond to a liver transplant. Chimeraplasty offers these children hope, and might reveal to be safer than the adenovirus approach, but the research is preliminary and the immune response is complex and mysterious.

Recent success was reported for treating children suffering from the severe immune disorder SCID type XI [208]. Gene therapy involves removing the bone marrow from infants, isolating their stem-cells, inserting the normal genes to replace the defective genes via retroviruses, and then re-infusing the stem cells into the blood stream. As hoped, the inserted stem cells were able to generate the cells needed for proper immune functioning in the patients, allowing the babies to live normal lives. Though successful for 2–3 years for most infants, complications arose when several infants developed leukemia-like conditions and one child even died. Scientists believe that the retrovirus vectors lodged near a cancer-causing gene and activated it. Therefore, alternative vectors for carrying the genes into the body have been under development, including the HIV virus, modified so it could not cause the disease. Clearly, weighing the overall benefits against the risks remains an issue for gene therapy. In addition, questions regarding the long-term behavior of the children’s new immune systems remain open.

Though clearly many bumps in the road are expected when new therapies are developed, scientists remain hopeful. Indeed, success in such gene therapy endeavors would lead to enormous progress in treating inherited diseases caused by point mutations.

2.4.5 Designed Compounds and Foods

From our farms to medicine cabinets to supermarket aisles, designer foods are big business.

As examples of these practical applications, consider the transgenic organisms designed to manufacture medically-important compounds: bacteria that produce human insulin, goats whose milk contains proteins to make silk for use in surgical thread or bulletproof clothing, silkworms that produce mammalian-type collagen and silk for use in tissue engineering and other medical applications, and the food product chymosin to make cheese, a substitute for the natural rennet enzyme traditionally extracted from cows’ stomachs. Genetically-modified bacteria, more generally, hold promise for administering drugs and vaccines more directly to the body (e.g., the gut) without the severe side effects of conventional therapies. For example, a strain of the harmless bacteria Lactococcus lactis modified to secrete the powerful anti-inflammatory protein interleukin-10 (IL-10) has shown to reduce bowel inflammation in mice afflicted with inflammatory bowel disease (IBD), a group of debilitating ailments that includes Crohn’s disease and ulcerative colitis.

The production of drugs in genetically-altered plants — “biopharming” or “molecular pharming” — represents a growing trend in agricultural biotechnology. The goal is to alter gene structure of plants so that medicines can be grown on the farm, such as to yield an edible vaccine from a potato plant against hepatitis B, or a useful antibody to be extracted from a tobacco plant.As in bioengineered foods, many obstacles must be overcome to make such technologies effective as medicines, environmentally safe, and economically profitable. Proponents of molecular pharming hope eventually for far cheaper and higher yielding drugs.

Genetically-engineered crops are also helping farmers and consumers by improving the taste and nutritional value of food, protecting crops from pests, and enhancing yields. Examples include the roughly one-half of the soybean and one-third of the corn grown in the United States, sturdier salad tomatoes,⁹ corn pollen that might damage monarch butterflies, papaya plants designed to withstand the papaya ringspot virus, and caffeine-free plants (missing the caffeine gene) that produce decaffeinated cups of java.

The general public (first in Europe and then in the United States) has resisted genetically-modified or biotech crops, and this was followed by several blockades of such foods by leading companies, as well as global biosafety accords to protect the environment. Protesters have painted these products as unnatural, hazardous, evil, and environmentally dangerous (‘Frankenfoods’).¹⁰

With the exception of transferred allergic sensitivities — as in Brazil nut allergies realized in soybeans that contained a gene from Brazil nuts — most negative reactions concerning food safety may not be scientifically well-grounded. In fact, not only do we abundantly use various sprays and chemicals to kill flies, bacteria, and other organisms in our surroundings and on the farm; each person consumes around 500,000 kilometers of DNA on an average day! Furthermore, there are many potential benefits from genetically-engineered foods, like higher nutrients and less dependency on pesticides, and these considerations might win in the long run. Still, environmental effects must be carefully monitored so that genetically-altered food will succeed in the long run (see Box 2.6 for possible problems).

Perhaps to counter fear of introduced allergens, bioengineering is also being used to reduce or remove compounds that cause allergic reactions in people. Though at a relatively early stage, various companies worldwide are using genetic engineering to try to reduce allergies from foods like wheat, rice, soybean, ryegrass, and peanuts. Genes responsible for producing allergenic proteins can be removed (i.e., knocked out), as done for soybeans, or the associated proteins redesigned, as in peanuts, so that allergenicity is lost but other nut characteristics are retained. As above, care must be taken to retain flavor, freshness, and looks of the original product, and not to introduce other possible allergens.

In addition to tampering with plants to remove allergens, such biotech companies are also expanding effort on the removal of genes associated with natural toxins. For example, companies (with support of national security organizations) are attempting to remove the toxin ricin —one of the deadliest substances known — from castor plants. Castor beans have been cultivated for centuries, and the plant’s natural oils (which lack toxicity) are widely used as laxatives and as component in brake fluid, dyes, soaps, and cosmetics. However, the toxic protein ricin can also be extracted from the castor plant, and has been associated with terrorist groups like Al Qaeda, with production of weapons “for mass destruction” in Iraq, and with an infamous killing of the spy Georgi Markov on a London sidewalk in 1978 by Bulgarian agents who injected ricin from an umbrella tip into the defector’s leg. Once removed, ricin-free castor plants can become more attractive to growers.

Box 2.6 Nutraceuticals Examples

The concept of fortified food is not new. Vitamin-D supplemented milk has eradicated rickets, and fortified breakfast cereals have saved many poor diets. In fact, classic bioengineering has been used for a long time to manipulate genes through conventional plant and animal inter-breeding. But the new claims — relying on our increased understanding of our body’s enzymes and many associated vital processes — have been making headlines. (“Stressed Out? Bad Knee? Try a Sip of These Juices.”, J.E. Barnes and G. Winter, New York Times, Business, 27 May 2001). Tea brews containing sedative roots like kava promise to tame tension and ease stress. Fruit-flavored tonics with added glucosamine (building block of cartilage) and calcium are claimed to soothe stiff knees of aging bodies. (See Chapter 3 on the fibrous protein collagen). Fiber-rich grains are now touted as heart-disease reducers, and fiber-rich foods have appeared in items well beyond cereals. Herb-coated snacks, like potato corn munchies coated with ginkgo biloba, are advertised as memory and alertness boosters.

With this growing trend of designer foods, the effect of these manipulations on our environment demands vigilant watch. This is because it is possible to create ‘super-resistant weeds’ or genetically-improved fish that win others in food or mate competitions. This potential danger emerges since, unlike conventional cross-breeding (e.g., producing a tangelo from a tangerine and grapefruit), genetic engineering can overcome the species barrier — by inserting nut genes in soybeans or fish genes in tomatoes, for example. This newer type of tinkering can have unexpected results in terms of toxins or allergens which, once released to the environment, cannot be stopped easily. For example, the first genetically-modified animal to reach American dinner plates is likely to be a genetically-altered salmon endowed with fortified genes that produce growth hormones, making the fish grow twice as fast as normal salmon. The effect of these endowed fish on the environment is yet unknown.

Popular examples of fortified food products with added vitamins and minerals (e.g., calcium and vitamin E) that also help protect against osteoporosis are orange juice, specialty eggs, and some vegetarian burritos. Other designer disease-fighting foods include drinks enriched with echinacea to combat colds; juices filled with amino acids and herbs claimed to boost muscle and brain function; margarines containing plant stanol esters (from soybean or pine trees) to fight heart disease and cancer (by blocking cholesterol absorption from the digestive tract), as well as green teas enriched with ginseng and other herbs; super-yogurts to enhance the immune system; and tofu and yams to combat hot flashes. Such functional foods are also touted to lower cholesterol, provide energy, fight off depression, or to protect against salmonella and E. coli poisoning (e.g., yogurt fortified with certain bacteria). Many other enriched food products are under design, for example fruit with increased vitamin C levels using a recently-isolated gene in strawberries (GalUR) that plays an important role in the production of vitamin C.

Will Ginkgo Biloba chips, Tension Tamer cocktails, or Quantum Punch juice become part of our daily diet (and medicine cabinet) in this millennium?

2.4.6 Nutrigenomics

Closer to the supermarket, one of the fastest growing categories of foods today is nutraceuticals (a.k.a. functional foods or pharmaceuticals), no longer relegated only to health-food stores. These foods are designed to improve our overall nutrition as well as to help ward off disease, from cancer prevention to improved brain function. See Box 2.6 for examples.

However, while nutraceuticals in general may characterize the many products that flood our supermarket aisles with health claims concerning enhanced cartilage support, cholesterol maintenance, relief of stress and tension, or maintenance of healthy lung function, the emerging field of nutrigenomics is a serious and well-grounded discipline. Nutrigenomics, at the interface of genomics, nutrition, and health, was made possible by recent developments in high-throughput transcriptomics, proteomics, and metabolomics technologies. Nutrigenomics integrates the genomics sciences with nutrition by studying how nature (the presence of particular genes or mutations) and nurture (our food intake, given environmental and behavioral factors) interact to manifest disease or protect us from it.

In its simplest form, diets low in certain proteins can be recommended for patients with phenylketonuria, or diets high in liver, broccoli, and other folic-acid rich foods can be a remedy for people with a genetic variation that produces a less efficient enzyme involved in processing folic acid. More generally, nutrition modifies the extent to which certain genes are expressed because macro-nutrients like proteins, micro-nutrients like vitamins, and naturally-occurring bioactive molecules like flavonoids regulate gene expression. Some of these compounds like resveratrol in red wine are ligands for transcription factors, and others like the natural amine nutrient choline — found in the lipids that make up cell membranes and in the neurotransmitter acetylcholine — alter signal transduction pathways and chromatin structure, thereby also affecting gene expression epigenetically. Because single nucleotide polymorphisms (SNPs) can alter gene functions, much of the focus in nutrigenomics has been on how the interaction of nutrients with SNPs increase or decrease disease risk.

Folate, for example, is among the nutrients critical to genome stability because it can cause DNA damage. More generally, key nutrients like folate, vitamin E, vitamin B₁₂, niacin, or calcium are associated with a reduction in DNA damage, while riboflavins and biotin tend to increase such damage. The familiar advice to lower fat intake and increase amounts of cruciferous vegetables can be rationalized by the lowering by these agents of oxidative DNA damage, which occurs from environmental factors like tobacco smoke and dietary factors like ultra high-fat diets. Thus, folate and other antioxidants and phytochemicals are recommended because they enhance DNA repair and reduce oxidative DNA damage. Such dietary modifications can help compensate for inherited mutations that may impair DNA damage repair. Because of this connection between DNA damage/repair and nutrition, some cancer researchers have become particularly interested in nutrigenomics.

In addition to cancer, diabetes, obesity, and cardiovascular disease have been researched in connection with food intake. Genetic susceptibility to these diseases (e.g., APOE-ε4 polymorphism, associated with elevated total cholesterol and increased risk of type-2 diabetes and Alzheimer’s disease) can be counteracted in part by dietary modifications that include plant-rich, high-fiber and low-fat diets in combination with regular exercise. Thus, nutrigenomics is leading to customized diet ingredients and supplements that are tailored to genetic variations, but the field is only beginning.

2.4.7 Designer Materials

New specialty materials are also being developed in industry with the needed thermochemistry, stereochemistry (e.g., compounds that bind to one chemical but not its mirror image), and kinetic properties. Examples are enzymes for manufacturing detergents, adhesives and coatings, photography film, or biosensors for explosives. Fullerene nanotubes (giant linear fullerene chains that can sustain enormous elastic deformations [1406]), formed from condensed carbon vapor, have many potential applications. These range from architectural components of bridges and buildings, cars, and airplanes to heavy-duty shock absorbers, to components of computer processors, scanning microscopes, and semiconductors.

Long buckyball nanotube fibers have even been proposed as elements of ‘elevators’ to space in the new millennium [1406]. These applications arise from their small size (their thickness is five orders of magnitude smaller than human hair), amazing electronic properties, and enormous mechanical strength of these polymers. In particular, these miniscule carbon molecules conduct heat much faster than silicon, and could therefore replace the silicon-based devices used in microelectronics, possibly overcoming current limitations of computer memory and speed. Far from science fiction, NASA scientists believe that the first space elevator, to carry cargo, might be built in the not-too-distant future.

2.4.8 Cosmeceuticals

Cosmeceutical companies are also rising — companies that specialize in design of cosmetics with bioactive ingredients (such as designer proteins and enzymes), including cosmetics that are individually customized (by pharmacogenomics) based on genetic markers,such as single nucleotide polymorphisms (SNPs).Most popular are products for sun or age-damaged skin containing alpha hydroxy acids (mainly glycolic and lactic acid), beta hydroxy acids (e.g., salicylic acid), and various derivatives of vitamin A or retinol (e.g., the tretinoin-containing Retin-A and Renova topical prescriptions). Besides reducing solar scars and wrinkling, products can also combat various skin diseases. Many of these compounds work by changing the metabolism of the epidermis, for example by increasing the rate of cell turnover, thereby enhancing exfoliation and the growth of new cells. New cosmeceuticals contain other antioxidants, analogues of various vitamins (A, D, and E), and antifungal agents.

The recent information gleaned from the Human Genome Project can help recognize changes that age and wrinkle skin tissue, or make hair or teeth gray. This in turn can lead to the application of functional genomics technology to develop agents that might help rejuvenate the skin, or color only target gray hair or tooth enamel. Computational methods have an important role in such developments by screening and optimizing designer peptides or proteins. Such biotechnology research to produce products for personal care will likely rise sharply in the coming years.