Skip Internal Navigation
In living organisms, recognition and self-assembly reactions are some of the most fundamental molecular processes responsible for, among others, folding, interactions, and aggregation of proteins and nucleic acids. We laid the foundation for our work by studying basic principles relating the structural organization of helical macromolecules, specifically DNA and collagen, to their interactions and biological function. We continue to build our understanding of DNA–DNA interactions, particularly recognition and pairing between intact, double-stranded DNAs with homologous sequences. However, we have shifted the major focus of our research to collagen and other extracellular matrix proteins and proteoglycans. We investigate their role in cancer, fibrosis, osteogenesis imperfecta, Ehlers-Danlos syndrome, chondrodysplasias, osteoporosis, and other diseases involving connective tissues. Together with other NICHD and extramural clinical scientists, we strive to improve our knowledge of the molecular mechanisms contributing to these diseases. We hope to use the knowledge gained for diagnostics, characterization, and treatment, bringing our expertise in physical biochemistry and theory to clinical research and practice.
Procollagen folding and its role in osteogenesis imperfecta (OI)
Type I collagen is a triple-helical protein that forms the stable matrix of bone, skin, and other tissues. In bone, it is produced by osteoblasts, the cells responsible for synthesis and mineralization of new bone material. Even though type I collagen is by far the most abundant protein in all vertebrates, its folding presents a challenge for cells. Type I procollagen precursor folds within the endoplasmic eeticulum (ER). Yet we discovered that in the absence of chaperones, in an ER-like environment at body temperature, the equilibrium state of procollagen is a random coil; triple helix folding becomes favorable only below 35°C. Cells have to use specialized chaperones to fold procollagen within the ER. Given the massive amounts of procollagen produced by osteoblasts, abnormal procollagen folding leads to ER stress and osteoblast malfunction, which is likely the reason why mutations in procollagen and its chaperones are responsible for over 90% of severe OI. A better understanding of the relationship between procollagen folding, ER stress response to its misfolding, resulting osteoblast malfunction, and OI severity may thus be essential for finding more effective OI treatments.
The most common mutations affecting procollagen folding are Gly substitutions in the obligatory (Gly-X-Y) n sequence of the triple helix. Our studies revealed that the effect of a Gly substitution on triple helix stability is determined entirely by the structural region within which the substitution is located, yet the rate of triple helix folding is also affected by the identity of the substituting residue. From analysis of published studies of various collagen-mimetic peptides and our own studies of collagens from OI patients with over 50 different Gly substitutions, we found that several structural regions within the triple helix may be responsible for distinct OI phenotypes. In particular, the first 85–90 amino acids at the N-terminal end of the triple helix form an "N-anchor" region with higher-than-average triple helix stability. Gly substitutions within this region disrupt the whole N-anchor, preventing normal cleavage of the adjacent N-propeptide. Incorporation of collagen with uncleaved N-propeptides into fibrils leads to hyperextensibility and joint laxity more characteristic of the Ehlers-Danlos syndrome (EDS). Disruptions of a stable "C-anchor" region at the other end of the molecule delay initiation of the triple helix folding, potentially explaining predominantly lethal outcomes of alpha1(I) Gly substitutions within this region.
Over the last several years, we have focused on another large region of predominantly lethal alpha1(I) and extremely severe alpha2(I) mutations, which surrounds the mammalian collagenase cleavage site. We found that this region contains two low-stability, flexible sub-regions separated by a "clamp" with higher triple helix stability. Disruptions of this clamp by Gly substitutions result in triple helix unwinding propagating through many adjacent Gly-X-Y triplets. Even some non-glycine mutations have a pronounced effect on the stability and folding of collagen within this region. We demonstrated, e.g., that Arg780 to Leu or Cys substitutions result in significant triple helix destabilization, potentially explaining more severe OI than with other, non-glycine missense mutations. Our data suggest that destabilizing mutations within this region may be particularly detrimental for procollagen folding, partly because of the low imino acid content and partly because of the lack of HSP47–binding sites (HSP47 binding is essential for stabilizing the triple helix during its folding).
Translational studies of patients with novel or unusual OI mutations
In addition to Gly substitutions, we characterized consequences of other mutations, most importantly deficiencies in ER chaperones involved in procollagen folding (1). In collaboration with scientists from the Bone and Extracellular Matrix Branch of NICHD (spb), we demonstrated the importance of a CRTAP/P3H1/CYPB complex for normal procollagen folding, assisting spb scientists in the discovery of three new forms of severe recessive OI associated with deficiencies in the cartilage-associated protein (CRTAP), prolyl-3-hydrohylase (P3H1), and cyclophilin B (CYPB). We also assisted spb scientists in the characterization of mutations in the C-propeptide cleavage site of type I procollagen, which lead to an unusual OI phenotype with high bone mass yet increased bone fragility (4).
Presently, we are focusing on OI caused by deficiency in Pigment Epithelium Derived Factor (PEDF). PEDF was initially described as an anti-angiogenesis factor secreted by pigment epithelium. Yet, genetic deficiency of both PEDF alleles leads to severe, progressively deforming OI (type VI) rather than to out-of-control angiogenesis, as reported several months ago. PEDF is a collagen-binding protein and a close evolutionary relative of HSP47, which is a collagen-specific ER chaperone. Nevertheless, PEDF does not appear to be directly involved in type I collagen biosynthesis. In collaboration with Frank Rauch, we investigated collagen synthesis, folding, and secretion by dermal fibroblasts from several OI patients with different PEDF deficiencies. Our data suggest that PEDF may function as a negative regulator of collagen fibrillogenesis, e.g., by preventing aberrant fibrillogenesis away from cell surfaces. We hypothesize that PEDF may also be involved in regulating osteoblast maturation and/or function, which is a subject of our ongoing studies.
Murine models of bone disorders
Murine models offer unique opportunities for the systematic study of the molecular mechanisms of bone disorders. We work with all three existing murine OI models, including the oim mouse, which has nonfunctional alpha2(I) chains, the Brtl mouse which has a knock-in G349C substitution in the alpha1(I) chain, and a mouse with a knock-in G610C substitution in the alpha2(I) chain. Our studies revealed reduced stability of type I collagen in G610C animals, enhanced stability in oim animals, and abnormal collagen-collagen interactions in both these models. However, in Brtl animals, which exhibit a more severe phenotype, we found no significant abnormalities in the stability or interactions of type I collagen. Instead, we discovered selective retention and intracellular degradation of molecules with a single mutant chain, resulting in osteoblast ER stress and malfunction. Based on our advances in high-definition microspectroscopic imaging technology, we found similar changes in the extracellular bone matrix in all these models despite very different defects in collagen biosynthesis. We observed less regular organization of the matrix, higher mineral content, and lower collagen content than in matched normal controls. Given the nature of mechanical stresses in bone, the latter abnormalities are likely to increase bone fragility, contributing to or determining the disease phenotype. At least in these three models, disease severity may be related to osteoblast malfunction, caused by abnormal folding and secretion of mutant collagen molecules, rather than to abnormal interactions of secreted molecules in the extracellular matrix. However, more studies are needed to verify our hypothesis.
In the past several years, we also investigated bone pathology in the caudal vertebrae of mice with bone tumors caused by defects in protein kinase A, a crucial protein in cAMP signaling. In addition to abnormal organization and mineralization of bone matrix, in some of the animals we found novel bone structures that had not been reported before. For instance, we observed free-standing cylindrical bones with osteon-like organization of lamellae and osteocytes but inverted mineralization pattern, highly mineralized central core, and decreasing mineralization away from the center. Improved understanding of these bone structures may not only clarify the role of cAMP signaling in bone but may also suggest new approaches to therapeutic manipulation of bone formation in skeletal dysplasias.
In addition to fundamental studies, mouse models also provide unique opportunities for developing novel treatments. In the last several years, we participated in testing gene and cell therapy strategies for OI treatment in the Brtl mouse model. In particular, in collaboration with Antonella Forlino, we found that 2% engrafted wild-type (WT) donor cells were responsible for synthesis of about 20% of bone matrix as well as dramatic improvements in the matrix composition and mineralization. In contrast, 2% engrafted Brtl cells did not produce any detectable (less than 1–2%) collagen matrix upon transplantation into WT animals. The results support our hypothesis of Brtl osteoblast malfunction as the primary cause of bone pathology and further demonstrate that the OI phenotype amelioration in Brtl animals transplanted with WT bone marrow resulted from the engrafted donor cells rather than transplantation artifacts. They also raise hopes for future clinical applications of bone marrow or BMSC transplantation in OI, given that even small engraftment appears to provide significant benefits.
Collagen matrix pathology in cancer and fibrosis
One of our other important advances in the past several years was the characterization of a collagenase-resistant, homotrimeric isoform of type I collagen and its potential role in cancer, fibrosis, and other disorders. The normal isoform of type I collagen is a heterotrimer of two alpha1(I) and one alpha2(I) chains. Homotrimers of three alpha1(I) chains were found in carcinomas and fibrotic tissues as well as in rare forms of OI and EDS associated with alpha2(I) chain deficiency. We found the homotrimers to be at least 5–10 times more resistant to cleavage by all mammalian collagenases than the heterotrimers, and we determined the molecular mechanism of this resistance. Our studies revealed that homotrimeric type I collagen is produced by cancer cells but not by cancer-associated fibroblasts, suggesting that cancer cells might utilize this collagen isoform for building collagenase-resistant tracks, supporting invasion through less resistant stroma.
More recently, we found that the collagenase resistance of homotrimeric type I collagen may play an important role in pathogenesis of glomerulosclerosis. Our study of oim mice with heterozygous and homozygous alpha2(I) chain deficiency as well as mice overexpressing TGF-beta in kidneys revealed sclerotic accumulation of the homotrimeric collagen in glomeruli. Analysis of the expression of collagen chains and collagenases in oim mice suggested that the pathology was caused by synthesis of the homotrimeric collagen by alpha2(I)–deficient glomerular cells and inability of the cells to cleave this collagen (5). Without the alpha2(I) chain deficiency, glomerular cells produce the homotrimeric collagen only upon injury or in response to altered cytokine environment, as indicated by previous reports and our analysis of the TGF-beta mice. We believe that the mechanism of subsequent sclerotic accumulation of the homotrimeric collagen is likely to be the same as in the alpha2(I)–deficient mice.
Further analysis of normal and fibrotic tissues from different patients (e.g., scleroderma, uterine fibroids, and pheochromocytomas) and mouse models (e.g., glomerular sclerosis and endocrine tumors) suggested that the synthesis of homotrimeric type I collagen is not a universal feature of all fibrotic lesions. Instead, our data indicate that the homotrimers are produced by undifferentiated, dedifferentiated, or transformed cells but not by normal or activated collagen-producing mesenchymal cells. Currently, we are investigating the regulation of the homotrimer synthesis by various cells and developing approaches to selectively target the homotrimers for potential diagnostic and treatment applications in cancer and fibrosis.
High-definition infrared and Raman micro-spectroscopy of tissues: cartilage pathology in diastrophic dysplasia
Label-free micro-spectroscopic infrared and Raman imaging of tissues and cell cultures provides important information about chemical composition, organization, and biological reactions inaccessible by more traditional histological techniques. However, applications of these promising technologies had been severely restricted by the limited accuracy of micro-spectroscopic imaging of soft, hydrated specimens under physiological conditions. Over the past several years, we resolved this problem by designing special specimen chambers with precise thermo-mechanical stabilization, reducing the light path instabilities and improving the spectral reproducibility by over two orders of magnitude.
In addition to the studies of bone mineralization discussed above, we applied this newly developed high-definition (HD) technology to the analysis of cartilage pathology in a mouse model of dyastrophic dysplasia (DTD). Undersulfation of cartilage proteoglycans caused by mutations in the SLC26A2 sulfate/chloride antiporter is believed to be responsible for severe progressive cartilage degradation and skeletal deformities in DTD patients. In DTD mice, net sulfation is only slightly reduced at birth and normalizes with age. However, DTD articular cartilage progressively degrades with age, and bones develop abnormally. Our HD–infrared imaging revealed strong chondroitin undersulfation only within narrow regions of the growth plate and periarticular region but nearly normal sulfation elsewhere in epiphyseal cartilage from newborn DTD animals. The undersulfation disrupted a thin layer of well-oriented collagen fibers at the articular surface, a layer that is normally present in newborn mice. The malformation of this layer, which protects cartilage from mechanical damage and synovial enzymes, may explain the progressive degradation of cartilage with animal age despite normalization of chondroitin sulfation. The undersulfation correlated with the rate of chondroitin synthesis, as measured by micro-autoradiography of radiolabeled sulfate incorporation into the cartilage, suggesting that the intracellular sulfate deficit is caused by accelerated chondroitin synthesis. Such a deficit may reduce sulfation of heparan sulfate and other minor species involved in cell signaling in the narrow zones crucial for cartilage development.
Theory and measurements of sequence-dependent DNA-DNA interactions
Interactions between double-stranded (duplex) DNA molecules are usually presumed to be independent of the DNA sequence because the nucleotides are buried inside the double helix and shielded by the highly charged sugar-phosphate backbone. However, accumulating experimental evidence suggests that the sequence-dependent structure of the sugar phosphate backbone might be important for DNA–DNA interactions. To account for this structure, we have, over the past decade, been developing a theory of electrostatic interactions between macromolecules with helical patterns of surface charges. The theory accounts for structure-specific DNA–DNA interactions that result from preferential juxtaposition of the negatively charged sugar phosphate backbone with positively charged counterions bound in grooves on the opposing molecule. The theory provides explanations for the observed counter-ion specificity of DNA condensation, changes in DNA structure upon aggregation, and multiple packing arrangements observed in DNA aggregates. The theoretical predictions are in good quantitative agreement with measured osmotic pressures in DNA aggregates. The theory also describes the observed positions and shapes of the observed peaks in x-ray diffraction patterns from hydrated DNA fibers (2).
Effects of DNA sequence on interactions between double helices predicted by this theory, e.g., direct recognition of sequence homology between 100 base pair (bp) or longer sequences, may have particularly important biological implications. The speed and accuracy of sequence homology recognition is crucial for DNA repair, preventing DNA lesions that lead to cell death and cancer. Our studies and other published reports indicate that local, transient pairing of homologous sequences in intact DNA molecules may precede double-strand breaks, further recognition by protein-covered single strands, and strand crossover. In late 2009, a single-molecule study was published that reported selective binding of 1–5 kb duplex DNA fragments to homologous regions on much longer molecules. Surprisingly, this study revealed stable pairing of DNA duplexes at ionic conditions at which DNA aggregation was never observed before and pairing of straight, parallel double helices appeared to be theoretically impossible.
To rationalize the latter observations, we revisited the theory, eliminating the simplifying assumption that DNA duplexes remain straight and parallel to each other when they form a stable pair. We found that DNA molecules tend to supercoil, forming a braid. During the last year, we completed and published a theory for the electrostatic energy of such braids, demonstrating that formation of stable braided pairs of homologous double helices might be possible in a much wider range of solutions than previously believed (3). Our predictions for the dependence of such pairing on counterions, salt concentration, and temperature match the experimental observations. Furthermore, the theory suggests possible interpretations for a number of other puzzling phenomena. For instance, electrostatic stabilization of left-handed braids predicted by this theory may be an important factor in explaining why hyperthermophilic bacteria and archea need reverse gyrases to promote left-handed supercoiling of circular DNA, which provides more stable conformation (essential for protecting the genome at temperatures above 100°C). We are presently conducting experiments designed to test some of these ideas.