2015 Group 1 Project

From CellBiology

Extracellular Matrix 2015 Projects: Group 1 | Group 2 | Group 3 | Group 4 | Group 5 | Group 6 | Group 7

Projects are now locked for final Assessment.

SLRPBanner 01.png

Note: as you read through this page, you'll notice that some terms are emboldened. This means that a definition for them is found in the glossary section at the bottom of the page, which can be accessed by clicking on the term. Only the first appearance of each term is emboldened.

Small Leucine-Rich Proteoglycans

Proteoglycans are a large and diverse family of biological molecules that play important roles in arrangement of the extracellular matrix (ECM). They consist of a protein core and one or more covalently linked glycosaminoglycan (GAG) chains, large polysaccharides characterised by the presence of repeating disaccharide units,[1] which constitute the vast majority of the volume of these molecules (~95%).[1]

The negative charge on GAG chains provide proteoglycans with their gel-forming function.

Even in small concentrations, proteoglycans produce a hydrophilic gel in the ECM as a result of the negative charge on the GAG side chains.[1] The gels formed by proteoglycans may vary in pore size, in some cases functioning as filters to regulate the movement of various elements of the ECM, and at other times acting as scaffolding for tissue development (reviewed in [2]). They also bind growth factors, guide cell migration, and even affect cell differentiation and arrangements.

As the topic of proteoglycans is incredibly vast and complex, this page will focus on a specific family of proteoglycans: the small leucine-rich proteoglycans (SLRPs). These molecules are characterised by an extensive set of leucine-rich repeats (LRRs), a structural motif that forms an α/β horseshoe fold within the protein.[3] SLRPs have a specialised role in the body which is not yet fully characterised. However, it is known that they are involved in pathological processes, including fibrosis and cancer cell growth (reviewed in [4]). This is a logical function of these molecules, given their role in regulation of cell regeneration.

Here, we will focus on four SLRPs in particular: decorin, biglycan, fibromodulin, and lumican, describing their synthesis, structure, function, and associated diseases. But first, some history...

History

Early Research

Discovery and research in the field of proteoglycans began in the mid-19th century, following a series of experiments isolating chondroitin sulfate (a prominent GAG) from cartilage tissue.[5] Over the following century, several key figures dominated proteoglycan research. Karl Meyer (1899-1990) was well-known for proteoglycan research, and was a world leader in hyaluronan synthesis specifically; Erik Jorpes (1894-1973) worked on heparan sulfate, and Hascall and Sadjera explored cartilage proteoglycans and developed numerous techniques to isolate GAGs.[2] Alongside this, Dorfman, Silbert and Lindahl performed studies on glycosaminoglycan biosynthesis.[2]

This graph represents the number of published scientific papers indexed by PubMed that respond to the search term "proteoglycans."

Advances in Technology and Discovery

The general structure of glycosaminoglycans (GAGs) were well-characterised by the 1950s, but were at the time called "mucopolysaccharides" and were thought to be an extracellular space-filling "ground substance."[5] This view of the functions of proteoglycans, while superseded in some dimensions, remains very much the general idea of the role of these crucial molecules.

Proteoglycan research advanced greatly in the 1990s with the development of advanced structure elucidation and other molecular biology techniques.[6] The majority of research until this point focused on structure, particularly of the GAG chain substituents rather than the protein core. This is because the strong charge on the GAG chain makes it easy to crystalise and isolate from the full proteoglycan molecule. At this stage, the full function of members of the proteoglycan family had yet to be fully characterised. Various structure/function correlation studies emerged in the first decade of the 21st century,[7], [8] and this led to an explosion in proteoglycan research.

Future Avenues for Research

The secrets of proteoglycans remain mostly hidden from view, and scientists continue to work desperately to apply our growing theoretical knowledge base to practical scenarios. In addition to advancing theoretical knowledge of the structure and function of proteoglycans, research into the role of proteoglycans in pathology and potential therapeutic applications are becoming increasingly popular. Because of the role of proteoglycans in structural arrangement of tissues, large therapeutic potential exists in the use of proteoglycans in regenerative medicine.[9], [10]

A recent article by Krencik et al. (2015) has indicated proteoglycan dysregulation in the astrocytes of individuals with Costello syndrome, a genetic disease characterised by mental retardation and musculoskeletal abnormalities.[11]

Proteoglycans are also relevant to the field of cancer research. There is emerging evidence (see right) that overexpression of certain SLRPs, most notably lumcorin, can produce an antitumorigenic effect, and may work to reduce rates of migration and metastasis in certain cancers (including melanoma).[12] This presents an exciting new avenue for research which may lead to improved therapeutic strategies.

Synthesis

Modulating substance Effect on transcription
Interleukin-1β Down-regulates general proteoglycan[13]
Tumor Necrosis Factor alpha Down-regulates general proteoglycan[13]
Transforming Growth Factor β Up-regulates biglycan, decorin, and fibromodulin [14], [15]
Dexamethasone Up-regulates decorin and down-regulates biglycan [16]
FGF2 Down-regulates biglycan[17]
Endoglin Down-regulates lumican[18]
GAG synthesis
Nucleotide consumption in the cytosol activates sugar and sulphates to create uridine diphosphate glucose-sugars (UDP-sugars) and 3'-phosphoadenosine 5'-phosphosulphate (PAPS). UDP-sugars and PAPS are transported via the endoplasmic reticulum to the Golgi apparatus, where individual GAGs are formed.[19] PAPS is the donor of sulphate to sulphotransferase in the Golgi and cytosol.
A schematic flowchart depicting proteoglycan synthesis.
Core Protein synthesis
Human SLRPs are encoded by 18 genes located on seven different chromosomes. The SLRP genes form chromosomal clusters, which suggest they were created via duplication of chromosomal segments.[20] Large clusters of four SLRP genes are located on chromosomes 9 and 12, with the remainder residing on chromosomes 1, 11 ,17 ,19 and X.[21] The process of protein transcription is the step in which substances can modulate SLRp production upregulate or downregulate SLRP synthesis. The strand of mRNA created during transcription is then transported out of the nucleus to be translated by ribosomes located on the rough endoplasmic reticulum.

19571367

The central leucine-rich domain of the core protein is synthesized by amino acid polymerisation at ribosomes. Translation is initiated by assembly of ribosomes around the mRNA strand. Amino acids bonded by tRNA are transported to the ribosome and are deposited if the tRNA anticodon forms complimentary base pairs via hydrogen bonds with a mRNA codon. This process continues with multiple amino acids being deposited until a stop codon halts the process and the newly formed polypeptide chain is released[22]. Post translation the protein core is transported to Golgi apparatus. The proteins are then glycosylated via enzyme reactions which cause covalent binding of a GAG chain (via a linker tetrasaccharide) to a serine residue in the protein core.[20]

Structure

Curved protein structure caused by LRRs

SLRPs have two structural components: a protein core and one or more linked glycosaminoglycan moieties. The core protein has a domain consisting a number of LRRs which gives central domain a structure reminiscent of a horseshoe-shaped solenoid. Parallel beta-sheets on the concave side of this structure are interwoven with alpha, polypropaline II and 3^10 helices on the convex side.[23] It been has accepted that each LRR has a characteristic 11 residue sequence (LXX-LXLXXNXL) which was first proposed in the review 'The leucine-rich repeat: a versatile binding motif' in 1994.[24] These are flanked by four conserved cysteine residues at the N-terminal and two cysteine residues at the C-terminal. The presence of disulphide caps protect the hydrophobic cores of the outer LRRs in the tertiary structure of the molecule.

Human SLRPs are currently sorted into five classes based on number of leucine rich repeats, N-terminal cysteine residue location, attachment of GAG chains and chromosomal exon location. This five class division is first mentioned in a 2008 review by Iozzo.[25] In more recent reviews and research articles, this system of classification is not applied in a particularly strict way (in recognition of the limitations inherent to such a system), and many members are classed as "non-canonical proteoglycans" because of apparent differences in amino acids flanking the N-terminals, as well as the absence of GAG chains. These are categorised as Class IV and V SLRPs. The reason that they are still classified as SLRPs despite lacking the crucial GAG component are the similarities in function, location and cysteine clustering around the central LRR domain.[26] Another key distinction is the presence of a specific C-terminal cysteine-rich capping motif called the "ear repeat" in canonical SLRPs.[27] The long penultimate LRR extends outwards from the convex side of the horseshoe fold of the SLRP and returns before the final LRR. It must be noted that most studies before 2000 count the central sections of LRRs while more recent studies also include the 2 LRRs flanking these central sections. This page has followed contemporary convention.

Structure of important Class I(Decorin), II(Fibromodulin) and III(PG-Lb). Note only central LRRs shown; the 2 flanking LRRs are not included.


Class I
Members of Class I are distinguished by the N-terminus sequence of CX3CXCX6C. Additionally the central domain have 12 LRRS, the molecule is flanked by four disulphide bonded conserved cysteine rich residues and GAG side shains consist of chondroitin and dermatan sulfate. The C-terminal domain of this class has not been investigated extensively though believed to have about 50 amino acid residues and two disulfide-linked cysteine residues separated by 32 amino acids. Typical members of class I include biglycan and decorin which share a 57% homology[28], [29] with a single GAG attached to decorin and two attached to biglycan, the core proteins of both originate off 8 exons. The initial section of the protein has a 16 amino acid signal peptide and 14-21 amino acid proptide this creates a sight for xylosyltranferase binding allowing initiation of GAG synthesis[28]. Nontypical members of Class I include asporin and ECM2. Asporin does not contain Ser-Gly dipeptides and lacks a flanking amino acid region. ECM2 is included in class I because of a 35% LRR homology with decorin and a similar genetic origin (chromosome 9).


Class II
Have a N-terminus sequence of CX3CXCX9C and GAG chains consisting of keratan sulfate and polyactosamine/unsulfated keratan. All of the members of Class II are encoded by only 3 exons with a single central exon encoding for the 12LRRs in all of its members. THe original members of Class II are fibromodulin and lumican.[30], [31] with being added in more recent studies keratocan, PRELP, and osteomodlin.[32], [33], [34]. All members display a sequence similarity ranging from 37% to 55%.
Class III
Members of class III have N-terminal cysteine rich cluster sequence of CX2CXCX6C and 8 LRRs. The three members of this class are PG-Lb, osteoglycin and opticin.[35] The overall structure is encoded on 7 exons for overall structure, with an additional 3 specifically for LRRs.
Class IV
Members of class IV have a N-terminal cysteine-rich cluster sequence of CX3CXCX6-17C and no attacted GAG chains. This absence of GAGs makes this more recently formed class non-canonical. It consists of chondroadherin,[36] nyctalopin,[37] and tsukushi,[26] all of which consist of 11 LRRs, distinguishing them from Class V (which has much longer repeat sequences).
Class V
The N-terminal Cysteine rich cluster sequence is CX3-4CXCX9X, with no GAG chains classifying them as non-canonical SLRPs.[26] Class V is made up of podocan and podocan-like protein 1 which has 20 and 21 LRRs respectively supporting the theory of chromosomal duplication segments. Sections of Class V SLRPs are homologous with the sequences present in Class I and II. Only two genes for coding podocan are present in chromosome 1 and those for podocan-like protein 1 are present on chromosome 9.[26]

Function

SLRPs have a variety of common functions which either occur concurrently but independently, or in conjunction to achieve the same function. As a group, SLRPs play significant roles in various biological processes that involve protein interaction within the organism. This group plays its role in tissue organisation by managing the process of cell growth and monitoring the maturation of specialized tissues. Other functions include modulation of growth factor activities, regulating collagen fibrillogenesis and skin tensile strength, affecting tumor cell growth/invasion and influencing the transparency of the cornea.[38] There is consequently an observable pattern in the distribution of SLRPs. They are typically found in the greatest concentrations in articular cartilage and tendons, but there are also specific areas such as skin, bone and the sclera and cornea of the eye where SLRPs with specialised functions may be found.[39]

Schematic diagram showing the role of SLRPs in fibrillogenesis.

The most significant and common of these functions is the participation of SLRPs when bound to Collagen Type I & II. The ability of SLRPs to readily bind collagen fibres are their primary form of interaction with cellular structures, leading to a regulatory function by manipulation of collagen fibres and their formation and assembly into fibrils. This process involves both the intracellular and extracellular matrices with transfer across the cell membrane allowing the interaction to occur. The SLRPs initiate and oversee the lengthening of fibrils by attaching them to one another, resulting in fibril growth.[40] This SLRP-mediated attachment of the fibrils allows them to evolve into eventual filaments at which point they detach. The SLRPs used throughout this process are released and recycled to begin the process again.

There are secondary functions that specifically relate to each SLRP, and are heavily dependent on their distribution. A common theme of secondary functions is interaction with the cell membrane and the integrated proteins and receptors found within it.

Decorin has a core protein which binds to collagen Type I and II, inhibiting the rate of fibrillogenesis. Decorin also controls the activity of multiple tyrosine kinase receptors. Decorin has the ability to bind to proteins that have high affinity to TGF-β1 which has a knock-on effect, as the decorin-TGF-β1 complex more strongly inhibits factors effecting osteoclast proliferation. Decorin also binds with a high affinity to the ligand of Insulin-like Growth Factor (IGF-1), as well as the IGF-1 receptor (IGF-1R).[41] This significant role implies the molecule in the pathology of cancer and several other chronic diseases. Decorin is found in the articular cartilage of synovial joints.[42] The ability of decorin to bind collagen is due to its shape, which is perfectly complementary to collagen fibres. This grants it efficiency in use, as well as the high specificity of its function.[43]

A working model summarizing the concept of biglycan-driven function.

Biglycan is a ligand of innate immunity receptors, and stimulates multiple inflammatory signalling pathways. It also binds specifically to fibrillar collagen and has been shown to affect tumor growth. While playing significant roles in each of these processes, biglycan is also a major contributor to the wound healing process after damage has occurred. It is a trigger molecule essential to initiating the molecular interactions integral to the healing process. It is therefore most commonly found in bone, tendon and cartilage where damage is most likely to occur.[44] Biglycan also has input in the inflammatory response through its multiple interactions with membrane-integrated receptors. It can induce crosstalk among various families of receptors and regulate the interactions by being allowed to permit or deny receptor activity. It also has the ability to mimic toll-like receptors, a group of central receptors regulating innate immunity.[45]

Postulated mechanisms of the action of lumican in modulating key cellular functions

Fibromodulin is involved in multiple processes with significant contributions to regulation of ECM assembly, determining cell progress and general cellular organisation. However, its primary purpose is in assisting in wound healing via interaction with type I and II collagen fibres, as well as mediating growth-factor binding.[46] Specifically, it has been shown to significantly promote fibroblast migration into the wound area, aiding timely wound closure and reduced scar formation. This is a mandatory requirement for the process of angiogenesis, which in turn is a strict prerequisite for progression of the wound healing process. Fibromodulin is typically found in higher concentrations in cartilage, tendon and sclera tissue, with lower concentrations detected in skin and bone.

The specialised function of lumican is cellular regulation as an initiating molecule that interacts with the cell plasma membrane and triggers a range of biological processes including cell motility, adhesion and signalling as well as apoptosis and inhibition of tumor progression. Lumican is also specifically and primarily found on the cornea of the eye but also in skin and cartilage.[47] Lumican plays a role in inflammation and wound healing. In inflammation, it localises macrophages to the site of corneal injury and recruits neutrophils. In wound healing, it decreases synthesis in scar tissue and mediates Fas-Fas ligand interactions.[48]

The Role of SLRPs in Disease

Collagen organisation has an extremely vital role in the functioning of many regions of the body, particularly those that have large amounts of ECM, including the musculoskeletal system (tendons, joints, ligaments and muscles), cardiovascular system, skin, the eyes and kidneys. Given the role of SLRPs in collagen binding, it is no surprise that SLRPs are extremely important in the functioning of different tissues.

Abnormalities in the processing and synthesis of SLRPs can produce small but fundamental changes to their structure, affecting their functions. This can result in detrimental effects on the different organ systems in the body. Consequently, research into this field over the past decade has focused on the genetic alterations affecting SLRPs, and their involvement in the pathophysiological features seen in a variety of human diseases.

Some examples of diseases that are affected by mutations in SLRPs include Ehlers-Danlos syndrome, a condition where skin hyperextensibility is a common phenotype, a result of a decorin deficiency (left image),[49] atherosclerosis (centre image),[50] and muscular dystrophy - a result of biglycan deficiencies (right image)[51].

Ehlers-danlos syndrome.pngAtherosclerosis.png Muscular dystrophy.png


Due to the large amount of information available, this page will focus on 2 of the body areas that are affected by abnormalities in SLRPs: the eyes and the musculoskeletal system.

SLRPs and Ocular Diseases

In the past ten years, research into this field has produced ample evidence to show that the majority of SLRP mutations have detrimental effects for the eyes, thus highlighting the central role played by SLRP mutation in ocular diseases.

Photobiomicroscopy showing a) healthy cornea and b) opacification of the cornea [52]

Congenital Corneal Stromal Dystrophy

Congenital corneal stromal dystrophy (CCSD) is a rare inherited disorder which is characterised by opacification of the cornea shortly after birth. The presence of the white deposits in the corneal stroma result in opaque appearance of the cornea (as shown in the figure on the right) which worsens with age.[53] Therefore, this disease progressively degrades the vision of the affected individual.

A frameshift mutation resulting in the truncation of decorin is responsible for this disease, with the deletion of the C-terminal 33 amino acid sequence which is vital for the structure of SLRPs occuring as a result.[54] Research suggests that this abnormal form of decorin binds to collagen in an atypical way, disrupting the generation and organisation of the corneal collagen fibrils, leading to the key pathological features observed in this disease. The precise arrangement of the collagen fibrils are key for corneal transparency. Thus, when the mutated form of decorin disrupts this arrangement, corneal clouding occurs (reviewed in [25]).

Flow diagram summarising the key pathophysiological events that occur in CCSD

Chen et al. (2011) highlight the additional roles of truncated decorin by affecting the ability of normal forms of decorin to interact with signalling receptors, causing the subsequent reduction of other SLRPs.[53] This imbalance in the expression of SLRPs further affects the generation and arrangement of fibrils in the cornea, adding to the detrimental effects on its transparency.

Complete Congenital Stationary Night Blindness

Dark adaptation curve of normal individuals and of individuals with complete congenital stationary night blindness[55]

Complete congenital stationary night blindness (also known as nyctalopia) an inherited, non-progressive ocular disease causing suboptimal nocturnal vision.[25] At present, genetic sequencing has identified 59 mutations in the NYX gene that encodes nyctalopin, [56] itself a SLRP which has an important role in the synaptic transmission between the depolarising bipolar cells and photoreceptors.[55]

Expression of nyctalopin is high in the inner and outer plexiform layer,[37] and is situated on the dendrites of depolarising bipolar cells.[56] It is suggested that nyctalopin interacts with proteins and channels (eg TRPM1 and GRM6) which are important in the retinal signalling pathway, thus, mutated forms of nyctalopin cause abnormal synaptic transmission in the retina,[56] principally affecting the rod photoreceptors response to light.

Individuals often show no deficit during normal lighting conditions, however, suffer from abnormal night vision. This is demonstrated in the diagram to the right which shows that normal patients have a biphasic curve consisting of an initial cone-mediated phase, followed by a rod-mediated phase, however individuals with the condition only show the initial cone-mediated phase. Additionally, it is common for individuals to suffer from other visual defects including high myopia, nystagmus and reduced visual acuity.[55]

SLRPs and musculoskeletal diseases

Through the use of immunohistochemistry and other localisation techniques, it is widely accepted that SLRPs are highly expressed and have important functional roles in most muscular and skeletal regions. The evidence that SLRPs play a fundamental role in numerous aspects of bone and muscle homeostasis corroborates with the fact that many degenerative bone diseases and muscular dystrophies prevalent in modern society are associated with abnormalities in SLRPs.

Osteoporosis

Osteoporosis is a highly prevalent, debilitating disease associated with aging and characterised by the reduction in bone mass to a pathological level (reviewed in [57]). There is ample evidence which shows the localisation of SLRPs to many skeletal areas where they are present to regulate different phases of the ‘bone life-cycle’ including, remodelling, osteogenesis and matrix & mineral deposition.[58] Individuals with osteoporosis commonly suffer from vertebral pain and an increased risk of bone fractures.

Biglycan knockout mice seem healthy and viable at birth but display reduced mass, mineral density and growth rate as they increase in age.[59] These are symptoms that strongly resemble that of an individual with osteoporosis. Additionally, subsequent analysis of these mice highlighted an important role for biglycan in the regulation of peak bone mass which explains the phenotypic characteristics seen, all of which are major risk factors in the pathophysiology of the disease.[60] Further experiments have shown that in biglycan knockout mice, levels of osteogenic precursor cells reduce much more quickly with age compared to wild type mice.[61]

The large influence that deficiencies in SLRPs have on the development of osteoporosis infers that there is the potential for a genetic component to have implications in the pathophysiology of the disease, however it is yet to be fully investigated.

Glossary

Term Meaning (underline indicates derivation of term)
Antitumorigenic Preventing tumor growth and development.
CCSD Congenital corneal stromal dystrophy; a disease involving increasing opacity of the corneal stroma.
Dexamethasone An anti-inflammatory and immunosuppressant drug of the glucocorticoid class.
ECM Extracellular matrix; the component of tissue found outside of the cells comprising that tissue.
GAG Glycosaminoglycan; large polysaccharides containing repeating disaccharide units.
Gal Galactose; a kind of monosaccharide sugar.
GlcA D-glucuronic acid; a carboxylic acid commonly found in proteoglycan carbohydrate chains.
GlcNAc D-glucosamine; the biochemical precursor of all nitrogenous sugars.
IdoA L-iduronic acid; a major component of some GAGs.
kDa Kilodalton; the atomic mass unit equal to 1.660x10-24 kilograms.
LRR Leucine Rich repeat; a protein structural motif that forms a horseshoe fold.
LXX-LXLXXNXL A sequence of amino acids found in the core protein moiety SLRPs, where: X = any amino-acid; L = leucince, isoleucine or valine; N = asparagine, cysteine or threonine.
PAPS 3'-phosphoadenosine 5'-phosphosulfate; a coenzyme in sulfotransferase reactions.
SLRP Small leucine-rich proteogylcan
solenoid A device consisting of a coil of copper wire turned many times around a core tube.
TGF-β Transforming growth factor beta; a cytokine involved in cell differentiation and proliferation.
UDP Uridine diphosphate; a nucleotide diphosphate. Essentially ADP with a uridine instead of an adenosine.

References

  1. 1.0 1.1 1.2 Bruce Alberts, Alexander Johnson, Julian Lewis, Martin Raff, Keith Roberts, Peter Walter Molecular Biology of the Cell: 4th Edition. New York: Garland Science, 2002. Accessed via: http://www.ncbi.nlm.nih.gov/books/NBK26810/
  2. 2.0 2.1 2.2 <pubmed>11292371</pubmed>
  3. <pubmed>2656687</pubmed>
  4. <pubmed>16046829</pubmed>
  5. 5.0 5.1 <pubmed>8298250</pubmed>
  6. <pubmed>10383378</pubmed>
  7. <pubmed>16547006</pubmed>
  8. <pubmed>15501918</pubmed>
  9. <pubmed>25900055</pubmed>
  10. <pubmed>25913849</pubmed>
  11. <pubmed>25947161</pubmed>
  12. <pubmed>24098450</pubmed>
  13. 13.0 13.1 <pubmed>7814444</pubmed>
  14. <pubmed>10625699</pubmed>
  15. <pubmed>12623288</pubmed>
  16. <pubmed>7988427</pubmed>
  17. <pubmed>21834878</pubmed>
  18. <pubmed>14996436</pubmed>
  19. <pubmed>3304145</pubmed>
  20. 20.0 20.1 <pubmed>11152695</pubmed>
  21. <pubmed>3934179</pubmed>
  22. <pubmed>19571367</pubmed>
  23. <pubmed>8535781</pubmed>
  24. <pubmed>7817399</pubmed>
  25. 25.0 25.1 25.2 <pubmed>18463092</pubmed>
  26. 26.0 26.1 26.2 26.3 <pubmed>12796502</pubmed>
  27. <pubmed>19077264</pubmed>
  28. 28.0 28.1 <pubmed>2647739</pubmed>
  29. <pubmed>3484330</pubmed>
  30. <pubmed>2531085</pubmed>
  31. <pubmed>1370446</pubmed>
  32. <pubmed>8621655</pubmed>
  33. <pubmed>7592739</pubmed>
  34. <pubmed>9642227</pubmed>
  35. <pubmed>1730648</pubmed>
  36. <pubmed>11445564</pubmed>
  37. 37.0 37.1 <pubmed>16553780</pubmed>
  38. <pubmed>15788868</pubmed>
  39. <pubmed>1740236</pubmed>
  40. <pubmed>25819456</pubmed>
  41. <pubmed>24389353</pubmed>
  42. <pubmed>8893759</pubmed>
  43. <pubmed>19753304</pubmed>
  44. <pubmed>24373744</pubmed>
  45. <pubmed>23350913</pubmed>
  46. <pubmed>25587509</pubmed>
  47. <pubmed>23438179</pubmed>
  48. <pubmed>24558602</pubmed>
  49. <pubmed>23587214</pubmed>
  50. <pubmed>23573226</pubmed>
  51. <pubmed>25055047</pubmed>
  52. <pubmed>9606218</pubmed>
  53. 53.0 53.1 <pubmed>21893019</pubmed>
  54. <pubmed>15671264</pubmed>
  55. 55.0 55.1 55.2 <pubmed>23658786</pubmed>
  56. 56.0 56.1 56.2 <pubmed>25802485</pubmed>
  57. <pubmed>13680312</pubmed>
  58. <pubmed>22879588</pubmed>
  59. <pubmed>9731537</pubmed>
  60. <pubmed>12213783</pubmed>
  61. <pubmed>11811564</pubmed>