Proline All-in-One MS-6657-W7HP
Proline ALL-IN-One Nanow are AP2020 All-in-One-PC, DC T4300 (2.1Ghz, 800Mhz FSB), MCP7A-LP, Nvidia ION VGA, 20" WXGA LCD 1600 x 900 5ms - M-Touch 2x 2GB DDRDimm, HD Audio, 10/100/1000 Lan, 640 GB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 6-in-1 card reader, DVD-RW Operating Softw are: Window s 7 Home Premium. Dim entions: 509.7 x 392.6 x 54.7mm. Matching Keyboard and Mouse included
MSI All-in-One MS-AE1920-B
MSI Wind Top AE1920 All-in-One-PC, Atom D525 (1.8Ghz), NM10, GMA3150 VGA, 18.5" WXGA LCD 1366 x 768 5ms - Single Touch 1x 2GB DDRDimm, HD Audio, 10/100/1000 Lan, 250 GB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 6-in-1 card reader, DVD-RW Operating Softw are: Window s 7 Home Premium. Dim entions: 476 x 368.6 x 49mm. Matching Keyboard and Mouse included MSI Wind Top AE2020 All-in-One-PC, DC T4500 (2.3Ghz, 800Mhz FSB), MCP7A-LP, Nvidia ION VGA, 20" WXGA LCD 1600 x 900 5ms - M-Touch 1x 2GB DDRDimm, HD Audio, 10/100/1000 Lan, 320 GB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 6-in-1 card reader, DVD-RW Operating Softw are: Window s 7 Home Premium. Dim entions: 509.7 x 392.6 x 54.7mm. Matching Keyboard and Mouse included MSI Wind Top AE2220 All-in-One-PC, DC T4500 (2.3Ghz, 800Mhz FSB), MCP7A-LP, Geforce 9300 VGA, 21.5" WXGA LCD 1920 x 1020 5ms - M-Touch 2x 2GB DDRDimm, 7.1 HD Audio, 10/100/1000 Lan, 640 GB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 4-in-1 card reader, DVD-RW Operating Softw are: Window s 7 Home Premium. Dim entions: 560 x 465 x 59mm. Matching Keyboard and Mouse included MSI Wind Top AE2260 All-in-One-PC, DC E5500 (2.8Ghz, 800Mhz FSB), G41, ATI 5430 VGA, 21.5" WXGA LCD 1920 x 1020 5ms - M-Touch 2x 2GB DDR3 Dimm, 7.1 HD Audio, 10/100/1000 Lan, 640 GB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 4-in-1 card reader, DVD-RW Operating Softw are: Window s 7 Home Premium 64. Dim entions: 555.1 X 403.93 X 75.42 mm. Matching Wireless Keyboard and Mouse included MSI Wind Top AE2420 All-in-One-PC, Core i(3.2Ghz) , H55, ATI HD5730, 23.6" WUXGA LCD 1920 x 1020 5ms - M-Touch, 2x 2GB DDR3 Dimm 7.1 TXT TrustAudio Pro, 10/100/1000 Lan, 1TB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 6-in-1 card reader, DVD-RW Operating Softw are: Window s 7 Home Premium 64. Dim entions: 617.28x461.76x71.95mm. Matching Wireless Keyboard and Mouse included MSI Wind Top AE24203D All-in-One-PC, Core i(3.2Ghz), H55, ATI HD5730, 23.6" WUXGA LCD 1920 x 1020 5ms - M-Touch, 2x 2GB DDR3 Dimm 7.1 TXT TrustAudio Pro, 10/100/1000 Lan, 1TB 3.5" HDD SATA, Built in Speakers, 1.3MP Webcam, 802.11 b/g/n, 6-in-1 card reader, Blue-Ray DVD Operating Softw are: Window s 7 Home Premium 64. Dim entions: 617.28x461.76x71.95mm. Matching Wireless Keyboard and Mouse included 1x set of 3DActive Shutter glasses included click here for more information on the MSI AIO Range MSI Bluetooth Class 2 Micro USB Dongle MSI Windtop Optional TV Tuner for AE2220/AE2260/AE2400 MSI Windtop Optional Wall Mount support Kit for AE20xx / AE22xx / AE22xx / AE24xx MSI AiO 3YR CARRY-IN (CLASSIC, GAMING, X-SLIM & ENTERTAINMENT) MSI AiO 3YR ONSITE (CLASSIC, GAMING, X-SLIM & ENTERTAINMENT)

MS-AE2020-B

MS-AE2220-B

MS-AE2260-B

MS-AE2420-W

MS-AE2420-B3D

Black/3D USB 3
Accessories MS-BTOES2 MS-AETUNER MS-AEWMB INFMSI3YCI-PC INFMSI3YOS-PC
Requires Technical installation Adds w all mount support/ w allmount brackets not included upgrades MSI AiO from 2 to 3 Years upgrades MSI AiO from 2 to 3 Years

The polyproline 11 helix Naturally, a Pro-Pro dipeptide is even more restricted, and a considerable body of evidence [10-14] suggests that a sequence of four or more proline residues in a row adopts a single preferred conformation in solution, with = -78 and Vf = + 146 , known as the polyproline II helix [15] (Figure 1). This is an extended structure with three residues per turn. It is found
Abbreviations used: PRR, proline-rich region; PRP, proline-rich protein.
Figure 1 Part of a polyproline 11 helix
The helix is extended and repeats every three residues. The proline 0 and ?4 angles are indicated. The 0 angle is constrained by the proline ring, while steric interactions between the proline a carbon and the preceding residue limit the conformational freedom of the preceding residue: if the preceding residue is in the a-helix conformation, the interactions drawn as dashed lines are energetically unfavourable. Nitrogen atoms are shown in pink, and oxygen atoms in red. Figure prepared using the program MOLSCRIPT [3].

M. P. Williamson

Table 1 Proteins with repetitive short proline-rich sequences
Name Light chain myosin kinase /?B1 crystallin OmpA Procyclin TonB
Source Rabbit skeletal muscle Ox eye lens E coli 1: brucei Bacterial

S. equi Streptococcus

Sequence

Protein function

Binds actin Cytoskeletal binding? Major outer membrane protein Membrane-bound coat protein Iron siderophore transport
Comment PRR is at N-terminus PRR is at N-terminus Mediates F-dependent conjugation Developmentally regulated Spans periplasmic space; binds FhuA Next to membrane anchor; antigenic 195 residues from membrane anchor

References

23 24,27

GP3GPAPGSG(PA)5Q(PA)2

(DP)2(EP)22-29

(EP)5X13(KP)5

(DPX)17 (15 are DPV) (XPZ)30 (X = T, S, A, I, L, V; Z is alternately + and -) PQP nine times

30 31-35

Group C M protein (equine) Group B IgA receptor
Binds peptidoglycan Binds peptidoglycan?
Outer membrane protein. Cell adhesion ? Tooth structure

p70 pertactin

Amelogenin

Bordetella parapertussis

PQP region not involved in adhesion
(QPX)9; 49 P in 170 residues
as Type VI turns [20].13C n.m.r. studies on proline-rich sequences in proteins show that cis Pro is rare in such sequences [21,22]. Following this brief review of proline conformation, we turn to consider proline-rich regions (PRRs) in proteins, concentrating on the large number of PRRs that contain repetitive proline-rich sequences, or multiple tandem repeats with minor variations between repeated sequences. In many of the examples discussed, the function of the PRR is uncertain. It is shown that the common element in almost all examples is that of binding, in a non-stoichiometric but functionally important way. However, in some cases, the PRR is largely used as a structural element; this function occurs most frequently in polypeptides containing hydroxyproline rather than proline. The division between the different sections below is intended to be by the type of sequence of the proline-rich section, although the differences occasionally become a little blurred.

Tandemly repeated sequences This group of proteins contains longer proline-rich sequences, typically 5-8 residues in length, which are repeated in tandem many times, often with slight variations (Table 2). In some cases, such as the salivary PRPs and the cereal storage proteins, the tandem repeats constitute almost the entire protein. The proteins
of this group have better characterized functions than most of the proteins discussed in the previous section; nearly all of the functions involve protein-protein binding. One of the best characterized groups is the salivary PRPs, which form 70 % of the protein in saliva. They appear to have several functions, but the most likely function of the proline-rich tandemly repeated section (which forms by far the largest part of the protein) is to bind plant polyphenols (tannins) present in the diet and to reduce their harmful effects by forming precipitates [51]. They do this by having long open extended structures which present a maximum surface area per residue, and achieve the precipitation of polyphenols by multivalent binding and noncovalent cross-linking [52], in much the same manner as multivalent antibodies bind, cross-link and precipitate antigens (Figure 4). The proline residues act not only to keep the structure

Proline-rich proteins

tannin-binding protein described above. They are large proteins, with a long tandemly repeated section. For example, the human tumour-associated polymorphic epithelial mucin has a 20-residue proline-rich sequence repeated between 21 and 125 times [41]. It is heavily glycosylated and is thought to function by creating an extensive network of interlocking extended chains anchored to the membrane, thus coating and lubricating the epithelial layer. However, its sequence similarity with the fungal protein suggests that it may have an additional function as a tannin-binding protein. The parasitic circumsporozoite protein is of particular medical interest because its repeated proline-rich sequence makes it highly immunogenic. Its function is to form a tough interlocking network, as does the dec- I eggshell protein (see below). The plant storage proteins play a vaguely analogous role, forming a tough extensible layer around the seed which is largely responsible for the texture of bread dough. The exact function of the plant storage proteins is unknown, but their situation in the periphery of the protein bodies [55] suggests that they may be involved in the support of these cellular organelles. This is presumably achieved by non-covalent interactions between protein chains (mediated in large part by the prolines), since there are few covalent cross-links. Elastin is probably somewhat different: its elasticity is thought to derive from the presence of repeated ,spirals, in which the regularly spaced proline residues play a key part by forming tight turns. The elastin structure is therefore one of the very few cases where the structural role of proline is to form turns, rather than to stabilize extended structure. Several actin-binding proteins with highly repetitive sequences were described in the previous section. Others have longer tandemly repeated sequences, such as the actin-binding protein from Dictyostelium discoideum. Other tandem proline-rich repeats are thought to be involved in structural organization, such as the C-terminal extension of squid rhodopsin. It is therefore clear that the longer tandemly repeated sequences discussed in this section play a qualitatively different role from that played by the (XP). and (XPY). sequences discussed in the previous section. Their greater length and flexibility allow them to form interlocking networks of high overall strength, suitable for external coats and irreversible precipitation of toxins. Nevertheless, the unique ability of PRRs to bind rapidly and tightly forms a common unifying motif.

Fe-ferrichrome

Outer membrane

To n f

Cell membrane

ADP A DI

Figure 3 Hypothetical model of the mechanism of TonB
The proline-rich section extends across the periplasmic space and provides a mechanical link between an intracellular ATPase and the siderophore transporter FhuA. It is proposed that the binding of an iron-ferrichrome complex to FhuA triggers a conformational change in TonB that activates the ATPase. The hydrolysis of ATP produces a further conformational change in TonB that opens the siderophore channel, allowing transport of the iron complex. Exchange of ADP to ATP completes the cycle.
Table 2 Proteins with tandemly repeated proline-rich sequences

Name Source

Comment
Most of the protein is PRR Glycosylated Between two small domains Small N- and C-terminal domains Small N- and C-terminal domains Small N- and C-terminal domains , spiral? Binds actin? at membrane? Organizes microvillar structure? Binds TFIID? Binds vesicle and cytoskeleton?

39,42 43

Salivary PRPs Mucins Circumsporozoite protein Gluten
C hordein Glutelin (zein) Elastin Actin-binding protein Rhodopsin RNA polymerase 11
Man, mouse Man Plasmodium berghei Wheat
(PQGPPQQGG), (GSTAPPAHGVTSAPDTRPAP),

(P4NPND)13PAPPQGN3(PQ)l7

GYYPTSPQQ, PGQGQQ; many repeats PQQPFPQQ many times VHLPPP eight times
Polyphenol binding Lubrication of epithelium Outer coat Cereal storage protein
Cereal storage protein Cereal storage protein Elastic connective tissue Actin assembly Vision Transcription Regulates vesicle release?

Barley

Maize Man
Dictyostelium discoideum Squid

Man Man

(VPGVG),

[GYP(P)Q(P)]5 (PPQGY)10

PQPAGPPAQQVPPPQQG ( x 3)

YSPTSPS (26 times)

Synapsin
Table 3 Non-repetitive PRRs
indicates a hydrophobic amino acid.

Source

Man Man Man Ox
PPHLNPQDPLKDLVSLACDPASQQPGPPTLRPTRPLQTVPLT PLPHFP2SLP2THSPTHP3AP3AP9 PPSGPAPDAQGGAPGQPTGPPGAPP PAVPPARPGSRGPAPGPPPAG PPLPPSTGRPAPAIPNRPGGGAPPLP XPXXPPP-XP

Protein tfunctionComment

59 60,61
CTF/NF-1 family Wilms tumour locus VAMP-1 Dynamin shibire gene product Consensus SH3-binding sequence mSosl Vitelline
Dec-1 eggshell protein Colostrum PRP

Transcription activator

Transcription activator Regulates vesicle release? Mediates early stages of receptor-mediated endocytosis Endocytosis

Transcription

Figure 5 Hypothetical model of the prelnllatlon complex of RNA polymerase 11
RNA polymerase 11 (Pol11I) is shown with a globular domain and an extended proline-rich C-terminal domain (CDo), to which transcriptional activators can bind in a conformationally ill-detined manner. The protein marked T represents the class of specitic polymerase 1l-associated proteins such as the TATA-binding element. The proline-rich C-terminal domain allows rapid binding of RNA polymerase I1to the transcriptional activators, correct bending of the DNA, and the formation of a tunctional preinitiation complex. Phosphorylation of the C-terminal domain leads to its dissociation from the transcriptional activators and the start of transcription. Adapted trom [56] and [79].
The second system that involves mutual interactions of several
synaptic vesicle-associated neuronal proteins, of which the best characterized is synapsin Synapsin I is prolinerich throughout, containing in particular a 17-residue prolinerich sequence that occurs three times (residues 436-452, 460-476 and 620-636). The synapsins are soluble proteins that bind to the outside of synaptic vesicles and probably also to the cytoskeletal matrix [82]. Phosphorylation of serines in the PRR near the C-terminus of synapsin I (Table 2) leads to a reduction in its binding to an incompletely characterized vesicle-associated

PRRs is the

protein [83], implying a role for synapsin Tin the phosphorylationdependent transition of synaptic vesicles from a 'reserve pooi' to a 'releasable pooi' of vesicles [84]. At least two other synaptic proteins, vesicle-associated membrane protein 1 (VAMP-i) and synaptophysin [85], contain proline-rich segments (Table 3). These are probably intrinsic membrane proteins with prolinerich cytoplasmic regions, which function by interacting with synapsin I, in this case as part of the system for activating synaptic vesicles for release. Other vesicle secretion and recycling systems appear to be
regulated by homologous proteins [86]. Of these, the best understood is dynamin (Table 3), which binds to an SH3 (src homology 3) domain, and is therefore discussed in the next section with other SH3-binding domains. Both the systems described in this section require the rapid and reversible association of several proteins into functional complexes, in which the prolines play a key part in the recognition and binding processes. Tight regulation of this association is necessary, which is achieved in both cases by phosphorylation of serines within the proline-rich sequence.

EGF receptor

InLzzz.Ras GDP
Non-repetitive PRRs Several other proteins function in similar ways to the tandemly repeated PRRs described above (i.e. by facilitating proteinprotein interactions), the only difference being that their PRRs are arranged in a non-repetitive manner. One such protein group that is currently of great interest is made up of the proteins that bind to SH3 domains (Table 3). SH3 domains are about 60 residues long, and have been found in association with catalytic domains, as in phospholipase Cy, within structural proteins such as spectrin and myosin (in which they may regulate the cytoskeleton), and in small adaptor proteins such as Sem-5, Crk, Drk and Grb2. These adaptor proteins have received close attention because of their role in what now seems to be an evolutionarily conserved signalling pathway, leading from receptor binding to the stimulation of Ras and the start of a kinase cascade (Figure 6) [87,88]. The adaptor proteins consist of an SH2 domain, which typically binds to a phosphorylated receptor, and two SH3 domains, which bind to proline-rich sequences on the nucleotidereleasing factor Sos (similar in sequence to yeast CDC25 [90]). The binding is a somewhat atypical PRR-binding event. Although, like other PRR binding, more than one PRR is required [63,88], here the sequence requirements are rather stringent. Sos proteins from different organisms have fairly long and variable PRRs, but there is a consensus binding sequence, which is XPXXPPPV1XPX (i1 indicates a hydrophobic residue), with prolines 2, 7 and 10 being essential [63]. Other residues, particularly 1 and 11, confer specificity on the binding [64]. A Plo sequence alone is incapable of binding. This means that each Sos binds with different affinities to SH3 domains from different sources [91]. In addition to their function in signal transduction, SH3 domain/PRR complexes also act as part of the vacuole sorting and receptor-mediated endocytosis pathways, which probably have many features in common with signal transduction. Thus it is now clear that one route of receptor-mediated endocytosis involves binding of the SH2 domains of phosphatidylinositol 3kinase to autophosphorylated receptors [92]. Ptdlns 3-kinase also has an SH3 domain, which binds to a PRR at the Cterminus of dynamin [61]. Dynamin is a GTP-binding protein, probably a GTPase, and shows sequence similarity to the Drosophila shibire gene product [62], mutants of which produce paralysis due to a defect in endocytosis. The N-terminal GTPbinding domain is similar in sequence to the yeast VPS1 (vacuolar protein sorting)/SPO15 gene product, which is involved both in vacuolar sorting and in meiotic chromosome segregation [93], while the C-terminal PRR is similar to the kinesin-related yeast KAR3 protein, which binds microtubules in vivo [94]. Dynamin is also thought to bind to microtubules. It is therefore tempting to postulate a pathway for endocytosis analogous to the signal transduction pathway shown in Figure 6, in which the chain of signal transduction proteins (phosphorylated receptor-Grb2mSosl-Ras) is replaced by the chain phosphorylated receptorPtdlns 3-kinase-dynamin-microtubule. However, both Ptdlns 3-

A/WVV\N

'zffiZZZ2Z.A22Z.Z.2

- - t//

ZffiffiZ/s -

Kinase cascade

Figure 6 Mechanism of the activation of Ras by the factor (EGF) receptor in mammals

epidermal growth

EGF binding leads to autophosphorylation of the tyrosine kinase (TK) domain of the receptor. The phosphorylated receptor then binds to the SH2 domain of the adaptor protein Grb2, which
produces a conformational change in its SH3 domains [89], allowing Grb2 in turn to bind to mSosl via two proline-rich regions with consensus sequence XPXXPPPiJXP (shown in red). mSosl is thought to act constitutively as a nucleotide exchanger, and its relocation to the plasma membrane activates Ras. The roles of EGF receptor/Grb2/mSosl in mammals are taken respectively by Sevenless/Drk/Sos in Drosophila, and by Let-23/Sem-5/unknown protein in Caenorhabditis elegans. Adapted from [87].
kinase and dynamin appear to have additional enzyme functions not possessed by Grb2 and mSosl. Several SH3 domain structures are now available [61,95,96]. In all cases the PRR-binding site is a smooth hydrophobic surface, rich in conserved aromatic amino acids, with charged amino acids at the periphery. It has been suggested [61] that the hydrophobic surface provides a general platform for binding the
the complex. The proline residues limit the conformational freedom of the linkers and prevent adjacent lipoyl domains from interacting with each other, which would reduce the enzymic efficiency of the complex [103]. The mixed alanine/proline sequence has more mobility than either an all-alanine or an all-
PRR, with selectivity resulting from the charged amino acids. The conformation of the bound PRR has not been determined, but from the shape of the binding site it seems likely to be an extended polyproline II helix. These results are consistent with the PRR acting as a 'sticky arm', binding rapidly and reversibly to SH3 domains. There are several proteins which are proline-rich throughout their entire sequence, notably the caseins and amelogenin. Caseins form about 800% of skim milk protein [97]. They have been divided by their electrophoretic mobility into as, ft. K and y caseins, constituting respectively about 50, 10, 30 and 5% of skim milk protein. They all contain prolines spread throughout the sequence in a fairly regular (but not repetitive) manner, with a representative bovine as casein having proline as 17 out of 186 residues, and a f casein A2 having 35 prolines out of 209 residues. The caseins assemble into micelles and clot by hydrolysis of a specific bond in K casein, catalysed by proteases present in the stomach [98]. The caseins are phosphorylated and bind calcium. The structure of the micelle is not clearly understood [99], but is apparently produced by a semi-ordered aggregation of core polymers formed by the association of extended polypeptide chains. The regularly spaced proline residues are presumably important in maintaining an extended chain conformation and also in guiding associative processes. Caseins, along with many other proteins, have sequences that are rich in proline, glutamic acid, serine and threonine and are flanked by positively charged residues. These sequences have been dubbed PEST sequences and have been suggested to be a signal for rapid degradation in eukaryotic cells, especially when phosphorylated [100,101]. The mechanism of degradation is as yet unknown, although it has been suggested that it may involve calcium-activated calpain proteolysis. The role of proline in the PEST sequence is unclear, but the similarities of this system to the phosphorylation-dependent RNA polymerase II and vesicleassociated protein systems may imply some evolutionary or functional similarities. Amelogenin (Table 1) is the predominant constituent of developing teeth. Bovine amelogenin has 170 residues and contains 49 prolines, of which nine form a (QPX)9 motif, with X = L, H or M [35]. Again, the protein functions by aggregation, and one can assume that the protein is largely extended, particularly the (QPX)9 motif, which is presumably an approximate polyproline II helix. Vitelline and the dec-l eggshell protein are involved in the strengthening of eggshell structure and therefore have similar functions to the tandemly repeated circumsporozoite protein described above. The other proteins listed in Table 3 have more poorly defined functions. A protein isolated from ovine colostrum is proline-rich throughout much of its sequence, and regulates the immune response, by binding in some way to surface receptors. The interdomain linkers in immunoglobulins, which constitute the main difference between different IgG subtypes, are rich in prolines. The function of the proline residues may be simply to maintain an extended structure with limited mobility. However, since the Fc receptor binding site is located close to the interdomain 'hinge', the prolines may also be involved in interactions with Fc receptors. There are a few other proline-rich sequences that appear to act solely as linkers, with no binding function. The most well studied is the approximately 30-residue linker that connects lipoyl domains in the dihydrolipoyl acetyltransferase component of 2-oxoacid dehydrogenase complexes [71]. This sequence contains essentially all-trans proline residues and is extended and mobile, even in the intact protein complex [102]. The function of the linker is to transfer acyl groups between different active sites in

proline sequence [104].

Table 3 ends with some striking polyproline sequences. In both Trypanosoma brucei protease and proacrosin these sequences appear to have multiple functions: interacting with other proteins, separating two domains and acting as cleavage sites after the protein has attached to its target. The remarkable polyproline sequences in papillomavirus and Epstein-Barr virus have as yet no known function, but they are likely to involve protein-protein association in a manner similar to the CTF/NF-I family of transcription factors. It is tempting to speculate that the Huntington's disease gene product is likewise involved in protein-protein association. Calcineurin A, like the Wilms tumour protein, contains a long stretch of continuous prolines, 11 in this case. It is one of the few proteins for which a specific (but highly speculative) model for PRP-protein interaction has been proposed. It is suggested that the Pi1 sequence, if in the form of a polyproline II helix, could extend along the calmodulin central helix, thereby presenting the few residues N-terminal to the P11 stretch in a fixed position on one calmodulin domain, determined by the calcineurin-binding site on the other domain. As we have seen, this requires an uncharacteristically specific interaction mode for the proline residues. In summary, all of the proteins presented in Tables 1-3 are very likely to have binding as a major function of the PRR, and in most cases binding is the only identifiable function. In a large number of cases the binding target is the cytoskeletal matrix, but many other ligands are also found. Proline-rich regions may therefore be taken to act as 'sticky arms' extending out from the rest of the protein (Figure 2). The binding is regulated by phosphorylation where required. The next section describes proline-rich regions that have a structural role rather than a role in binding. This change in function is achieved by hydroxylation of some or all of the proline residues.
Hydroxyproline-rich proteins
The most well-known PRP, and probably the most abundant in the animal kingdom, is collagen (Table 4). It is a stiff high-tensile fibre found in connective tissue such as tendons and skin, and comprises up to a third of total body protein. Three chains are coiled around each other to give a triple-stranded helix, which is stabilized by hydrogen bonding of glycine between strands. Each strand forms a polyproline II helix. The regular sequence is crucial for maintaining the collagen structure; models of the collagen structure show that substitutions by other residues lead to steric clashes or unpaired hydrogen bonds. The extended terminus of the blood complement protein Clq also seems to be collagen-like. It associates to form a triple helix, while a break in the (GXX). sequence at around residue 39 forces the individual chains to bend, forming a 'bunch of tulips'

structure.

The role of extensins is not clear.
plant cell wall, and probably strengthen it by covalent crosslinking of tyrosine residues. However, it is likely that the initial

They form 5-10 % of the

structure is produced by interwoven and non-covalently associated extensin chains. Extensins accumulate in plant cell walls upon wounding [113] and pathogen attack [114], indicating an
Table 4 Proteins rich in hydroxyproline
In this table, P denotes hydroxyproline
Source Man Man Tomato, carrot etc. Sorghum Cucumber

Volvox carteri

Protein function Stiff connective fibre Blood defence system Cell wall constituent Cell wall constituent Electron transfer

107-111

Collagen Clq Extensin P1 Hydroxyproline-rich glycoprotein Cucumber peel cupredoxin
(GPI)350 (with variations on P and P)
(PGX)3(ASXGX)2(FGX)2PGXP [SPPPP(VKPYHP)TPIKY]n (PATKPPTPPVYTPSPKP)n PPPSSSPFSSVMPPP/MPPPSPS
Triple helix Kinked triple helix Protects plant against damage? Many are hydroxylated
PRR is C-terminal extension (locates in matrix?) Also role in tannin binding?
Extracellular matrix protein
P2SP3SPfRP2SP4SFSP17SP18SFSP2

Strengthens cell matrix

additional role in defence, possibly due to agglutination of invading bacteria ([115] and refs. cited therein). This function is reminiscent of that carried out by the salivary PRPs, which also agglutinate bacteria. Many microbial polysaccharide-digesting enzymes consist of two domains, a catalytic domain and a sugar-binding domain. The two domains are separated by a semi-rigid linker, whose function appears to be largely to hold the two domains apart, although it may play some role in stabilization of the domains against heat or chemical denaturation [116]. The linker can have a wide variety of sequences (reviewed for ,-1,4-glycanases in [117]), which are generally rich in hydroxyamino acids (serine and threonine) or proline, or both. In the linkers of fungal proteins the hydroxyamino acids are heavily glycosylated, which both rigidifies and protects the linker [118]. The PRR of the extracellular matrix protein from Volvox carteri may play a similar role. It is possible that the PRR may also be involved in protection of the cell surface, by analogy with the epithelial mucins discussed above.
PROLINE IS INVOLVED IN BINDING
In this review I have sought to demonstrate that proline does not merely act as a spacer, but frequently has an important role in binding as well. This is true both for the (XP). sequences and for the longer more varied tandem repeats. Clearly, the binding generated cannot be highly specific, but it can be both very rapid (because of the small surface area and flexibility involved) and remarkably strong. Less specific binding can be of positive advantage in some cases, allowing a wider range of ligands to be bound. This is of relevance to salivary PRPs, which have to bind a wide range of polyphenols and other substrates, and possibly also for the transcription factors, which probably need to bind to a range of different proteins involved in the initiation of transcription. Commenting on these systems, Sigler [79] writes: 'These systems share the need for a mechanism by which many and various proteins can interact with a common cellular element. These flexible and variable contact patterns depart from the traditional view of specific molecular interactions gained from studying assemblies of globular molecules that give crystalline images'. This comment has been fully borne out by the fuller and more recent data reported here. The strength of the binding derives from the fact that prolinerich polypeptides have highly restricted mobility (and therefore relatively low entropy) even before binding. Thus binding leads to a smaller drop in entropy than it would do for a normal, more flexible, peptide, and hence a greater overall binding energy is achieved. To take an example, if we assume that each dipeptide

Xaa-Pro has only two rather than the normal four degrees of rotational freedom around the backbone bonds, and we further assume that on binding all rotational freedom is lost, then an Xaa-Pro dipeptide loses two fewer degrees of freedom on binding. It has been estimated [1 19] that each degree of rotational freedom is worth 5-7 kJ mol-1 at 300 K; more recent estimates [120] place the figure somewhat lower, at around 3.5 kJ * molP. Therefore the AG for the binding of an octapeptide (i.e. four dipeptides) could increase by 14 kJ mol-h, increasing the association constant from (for example) 103 to 2.7 x 105 M-1, a value approaching a reasonable number for specific binding (cf. values of 105-107 M-1 for peptides binding to major histocompatibility complex class I molecules of appropriate specificity [121]). The multiple tandem repeats often found in PRRs appear to be another device for increasing weak binding, in much the same way as divalent antibodies bind to antigens much more strongly than monovalent ones [52]. Thus, for example, salivary PRPs act to bind to and precipitate dietary polyphenols. The precipitation reaction is mediated by cross-linking of one PRP to several polyphenols, and of one polyphenol to several PRPs. As discussed above, PRRs often have the additional function of a structural element, and multiple repeats are also necessary for extending the length of the protein. For example, they provide the coccal cell wall proteins with the length to span the peptidoglycan layer, and incidentally thereby take full advantage of it for binding purposes. Similar observations have been made for so-called protein modules [122], which are single small protein domains that are repeated many times and which occur in a wide range of vertebrate blood and cell-surface receptor proteins. For example, fibronectin consists of 29 similar modules in a single polypeptide chain. The major function of these modular proteins appears to be in protein-protein binding, but they probably have an additional function of spacers, separating one functional part of the protein from another. Their functions therefore closely parallel those of the tandem proline-rich repeats discussed here. The non-specific nature of the binding is supported by evidence showing that the exact number or sequence of the proline-rich repeats makes little difference to protein function. Thus deletion of almost half of the repeats in the C-terminal domain of RNA polymerase II still produces more or less functional proteins [123].

THE NATURE OF THE BINDING INTERACTION Apart from the entropy advantage, proline has other features that make it a good ligand. It has a large flat hydrophobic surface and therefore binds well to other flat hydrophobic surfaces such as aromatic rings. Indeed, it is of more than passing interest that

F ree peptide

Complex
mone receptor [128], which has a P15 stretch immediately followed by a Q6 stretch; and the Huntington's gene product (Table 3). It has been suggested [129] that 'Q-linkers' form an identifiable class of interdomain linkers in multidomain regulatory proteins. Q-linkers share a number of characteristics in common with proline-rich linkers, and many proline-rich linkers also contain high proportions of glutamine, for example the salivary PRPs and the cereal storage proteins. The significance of this is unclear, but it may be relevant to note that the second most likely residue to appear in a polyproline helix segment in globular proteins is glutamine (proline being the most likely) [18]. Indeed, glutamine is the only residue other than proline to have a Chou-Fasman conformational parameter that is higher for a polyproline II helix than for any other category of secondary structure. Thus it may be that glutamine is preferred as a linker component because, like proline, it preferentially forms polyproline II helices and makes an extended, conformationally restricted, polypeptide

chain.

Complexation
Figure 7 Schematic free energy diagram for the binding of a proline-rich peptide (red) or a normal peptide (black) to a globular protein
Because the free PRP is less flexible, it has less entropy and therefore a greater free energy than the normal peptide, by an amount AGs. The bound peptides have similar entropy, but the PRP binds with more favourable enthalpy (heat of binding), by an amount AGH, because of its more electron-rich amide bond. The overall binding energy of the PRP, AGPRP, is therefore more favourable than that of the normal peptide, AGaorrmi, by AGs+AGH.

CONCLUDING REMARKS

one of the very few proteins shown to be a receptor for a PRP, the SH3 domain, has a binding site lined with conserved aromatic residues [95]. The salivary PRPs have also been shown to interact with their principal physiological target, polyphenols, via proline residues [53]. The crystal structure of avian pancreatic polypeptide [16] shows that the polyproline structure is stabilized by interactions between the prolines in the N-terminal polyproline II helix and non-polar side-chains (many of them aromatic) in the C-terminal a-helix. Neuropeptide Y [17] is stabilized by similar interactions. Although proline cannot act as a hydrogen bond donor, it is a very good hydrogen bond acceptor, possibly because the electron-donating potential of the methylene group attached to the amide nitrogen causes the amide carbonyl to be electron-rich [15,52,124]. It is presumably this property that causes prolinerich peptides to be highly soluble in water, and leads to confusion as to whether proline should be classed as a hydrophobic or a hydrophilic residue. Moreover, thermodynamic studies [125,126] have shown that tertiary amides such as the Xaa-Pro peptide bond are preferred to secondary amides as hydrogen bond donors, the enthalpy (heat) change for tertiary amides on forming a hydrogen bond being about 50 % more favourable than for secondary amides. This observation may well be one aspect of a more general phenomenon, namely that if a solute is well solvated, its tendency to interact with other species will be reduced. Proline is more poorly solvated than other amino acids, in that it is hydrated by fewer water molecules around the amide bond, and so interacts more strongly with other solutes. In summary, proline may well be a preferred ligand enthalpically as well as entropically (Figure 7). It may be of significance that many proline-rich sequences also contain large numbers of glutamine residues. Particularly striking are the nuclear protein SNF5, a transcription activator [127], which is proline-rich but also contains the sequence Q7HQ37; a Drosophila 20-hydroxyecdysone-inducible steroid nuclear hor-

6 Hurley, J. H., Mason, D. A. and Matthews, B. W. (1992) Biopolymers 32, 1443-Williams, K. A. and Deber, C. M. (1991) Biochemistry 30, 8919-Sankararamakrishnan, R. and Vishveshwara, S. (1993) Proteins Struct. Funct. Genet. 15, 26-Richardson, J. S. and Richardson, D. C. (1988) Science 240, 1648-Deber, C. M., Bovey, F. A., Carver, J. P. and Blout, E. R. (1970) J. Am. Chem. Soc. 92, 6191-Helbecque, N. and Loucheux-Lefebvre, M. H. (1982) Int. J. Peptide Protein Res. 19, 94-Okabayashi, H., Isemura, T. and Sakakibara, S. (1968) Biopolymers 6, 323-Dukor, R. K. and Kiederling, T. A. (1991) Biopolymers 31, 1747-Dukor, R. K., Kiederling, T. A. and Gut, V. (1991) Int. J. Peptide Protein Res. 38, 198-Cowan, P. M. and McGavin, S. (1955) Nature (London) 176, 501-503. 16 Blundell, T. L., Pitts, J. E., Tickle, I. J., Wood, S. P. and Wu, C.-W. (1981) Proc. Natl. Acad. Sci. U.S.A. 78, 4175-Darbon, H., Bernassau, J.-M., Deleuze, C., Chenu, J., Roussel, A. and Cambillau, C. (1992) Eur. J. Biochem. 209, 765-Adzhubei, A. A. and Sternberg, M. J. E. (1993) J. Mol. Biol. 229, 472-Grathwohl, C. and Wuthrich, K. (1976) Biopolymers 15, 2025-Lewis, P. N., Momany, F. A. and Scheraga, H. A. (1973) Biochim. Biophys. Acta 303, 211-Green, J. D. F., Perham, R. N., Ullrich, S. J. and Appella, E. (1992) J. Biol. Chem. 267, 23484-Bhandari, D. G., Levine, B. A. and Yeadon, M. E. (1986) Eur. J. Biochem. 160, 349-Frank, G. and Weeds, A. G. (1974) Eur. J. Biochem. 44, 317-Hejtmancik, J. F., Thompson, M. A., Wistow, G. and Piatigorsky, J. (1986) J. Biol. Chem. 261, 982-Berbers, G. A. M., Hoekman, W. A., Bloemendal, H., de Jong, W. W., Kleinschmidt, T. and Braunitzer, G. (1983) FEBS Lett. 161, 225-Chen, R., Schmidmayr, W., Kramer, C., Chen-Schmeisser, U. and Henning, U. (1980) Proc. Natl. Acad. Sci. U.S.A. 77, 4592-Roditi, I., Schwarz, H., Pearson, T. W., Beecroft, R. P., Liu, M. K., Richardson, J. P., BUhring, H.-J., Pleiss, J., Bulow, R., Williams, R. 0. and Overath, P. (1989) J. Cell Biol. 108, 737-Gaisser, S. and Braun, V. (1991) Mol. Microbiol. 5, 2777-Brewer, S., Tolley, M., Trayer, I. P., Barr, G. C., Dorman, C. J., Hannavy, K., Higgins, C. F., Evans, J. S., Levine, B. A. and Wormald, M. R. (1990) J. Mol. Biol. 216, 883-Timoney, J. F., Muktar, M. and Ding, J. (1991) in Genetics and Molecular Biology of Streptococci, Lactococci and Enterococci (Dunny, G. M., Cleary, P. P. and McKay, L. L., eds.), pp. 160-164, American Society of Microbiology, Washington 31 Hed6n, L.-O., Frithz, E. and Lindahl, G. (1991) Eur. J. Immunol. 21, 1481-Pancholi, V. and Fischetti, V. A. (1988) J. Bacteriol. 170, 2618-Fahnestock, S. R., Alexander, P., Nagle, J. and Filpula, D. (1986) J. Bacteriol. 167, 870-Li, L. J., Dougan, G., Novotny, P. and Charles, I. G. (1991) Mol. Microbiol. 5, 409-Takagi, T., Suzuki, M., Baba, T., Minegishi, K. and Sasaki, S. (1984) Biochem. Biophys. Res. Commun. 121, 592-Abillon, E., Bremier, L. and Cardinaud, R. (1990) Biochim. Biophys. Acta 1037, 394-Hannavy, K., Barr, G. C., Dorman, C. J., Adamson, J., Mazengera, L. R., Gallagher, M. P., Evans, J. S., Levine, B. A., Trayer, I. P. and Higgins, C. F. (1990) J. Mol. Biol. 216, 897-Fischetti, V. A., Pancholi, V. and Schneewind, 0. (1991) in Genetics and Molecular Biology of Streptococci, Lactococci and Enterococci (Dunny, G. M., Cleary, P. P. and McKay, L. L., eds.), pp. 290-294, American Society of Microbiology, Washington 39 Bennick, A. (1982) Mol. Cell. Biochem. 45, 83-Laytield, R., Bannister, A. J., Pierce, E. J. and McDonald, C. J. (1992) Eur. J. Biochem. 204, 591-Gendler, S. J., Lancaster, C. A., Taylor-Papadimitriou, J., Duhig, T., Peat, N., Burchell, J., Pemberton, L., Lalani, E. and Wilson, D. (1990) J. Biol. Chem. 265, 15286-42 Eichinger, D. J., Arnot, D. E., Tam, J. P., Nussenzweig, V. and Enea, V. (1986) Mol. Cell Biol. 6, 3965-Field, J. M., Tatham, A. S. and Shewry, P. R. (1987) Biochem. J. 247, 215-Tatham, A. S., Drake, A. F. and Shewry, P. R. (1985) Biochem. J. 226, 557-Pons, M., Feliz, M., Celma, C. and Giralt, M. (1987) Magn. Reson. Chem. 25, 402-Venkatachalam, C. M. and Urry, D. W. (1981) Macromolecules 14, 1225-1229