Opening Amyloid-Windows to the Secondary Structure of Proteins The Amyloidogenecity Increases Tenfold Inside Beta-Sheets

2025-04-27 1 0 1.18MB 12 页 10玖币

侵权投诉

Opening Amyloid-Windows to the Secondary Structure

of Proteins: The Amyloidogenecity Increases Tenfold

Inside Beta-Sheets

Krist´of Tak´acsa, B´alint Vargaa, Viktor Farkasc, Andr´as Perczelc,d, Vince

Grolmusza,b,∗

aPIT Bioinformatics Group, E¨

otv¨

os University, H-1117 Budapest, Hungary

bUratim Ltd., H-1118 Budapest, Hungary

cELKH-ELTE Protein Modeling Research Group, H-1117 Budapest, Hungary

dLaboratory of Structural Chemistry and Biology, E¨

otv¨

os University, H-1117, Budapest,

Hungary

Abstract

Methods from artiﬁcial intelligence (AI), in general, and machine learning, in

particular, have kept conquering new territories in numerous areas of science.

Most of the applications of these techniques are restricted to the classiﬁcation

of large data sets, but new scientiﬁc knowledge can seldom be inferred from

these tools. Here we show that an AI-based amyloidogenecity predictor can

strongly diﬀerentiate the border- and the internal hexamers of β-pleated sheets

when screening all the Protein Data Bank-deposited homology-ﬁltered protein

structures. Our main result shows that more than 30% of internal hexamers of β

sheets are predicted to be amyloidogenic, while just outside the border regions,

only 3% are predicted as such. This result may elucidate a general protection

mechanism of proteins against turning into amyloids: if the borders of β-sheets

were amyloidogenic, then the whole βsheet could turn more easily into an

insoluble amyloid-structure, characterized by periodically repeated parallel β-

sheets. We also present that no analogous phenomenon exists on the borders of

α-helices or randomly chosen subsequences of the studied protein structures.

Introduction

The amyloid status of proteins are studied for a long time because of its

relevance in human diseases, including neurodegenerative ailments [1, 2, 3, 4],

amyloidoses [5] and other diseases, listed in Table 1 of [2]. Recently, the amyloids

gained increased attention as possible antiviral agents [6], or in one of the ﬁrst

successes in the human in vivo CRISPR-Cas9-based gene editing therapy of the

transthyretin amyloidosis [7].

∗Corresponding author

arXiv:2210.11842v1 [q-bio.BM] 21 Oct 2022

The insoluble amyloid state diﬀers from the unstructured protein aggregates;

it has a well-deﬁned structure characterized by repetitive parallel β-sheets [8, 9].

Numerous globular proteins may turn into insoluble amyloid structures in

certain pH and temperature combinations. It is observed that the transition to

the amyloid state is similar to the contagious prion-formation: it is related to

a core-forming structure [2] for starting the amyloid-transition, and the subse-

quent propagation of misfolded, insoluble, parallel β-sheets [10, 11]. The amy-

loid propagation is related to the spatially accessible, neighboring β-pleated

sheet secondary structural elements of proteins [12, 3, 4].

Naturally occurring, globular, soluble proteins need to possess certain struc-

tural properties which prevent them from turning into insoluble amyloid aggre-

gates. Experimental studies, together with in silico methods able to predict

aggregation-prone regions (APRs) in protein sequences. In the literature, cer-

tain residues acting as “gatekeepers” are described as preventing those confor-

mational changes to amyloid structures: aspartic acid and glycine in bacterial

curlin subunits CsgA and CsgB [13], lysine in transthyretin [14] or in peptides

Trpzip1 and Trpzip2 [15].

Our hypothesis is that the borders of β-sheets in globular, soluble proteins in

general (and not only in the speciﬁc examples listed above) need to be protected

against turning to amyloid structures; otherwise the whole polypeptide chain

bordering the β-sheet subsequences would be transformed to amyloids. We

examine this hypothesis in the present work by screening the starting and ending

subsequences, or in other terminology, the preﬁxes, and suﬃxes of the β-sheets

of the homology-ﬁltered Protein Data Bank (PDB) entries.

Previous work

In [8], a clean geometrical formulation was given for the characterization

of the amyloid-like structures in length, distance, and parallelism of their β-

sheets, and by the search of these geometric properties, an automatically up-

dated list of the PDB-entries was published and maintained at the address

https://pitgroup.org/amyloid/.

In [9], the amyloid-precursor molecules of https://pitgroup.org/amyloid/

were classiﬁed by the preﬁxes of their parallel β-sheet subsequences, and these

preﬁxes were also searched for in the whole Protein Data Bank [16]. The remark-

able ﬁndings include the major histocompatibility complexes MHC-1 and MHC-

2, the p53 tumor suppressor protein, and an anti-coagulant peptide molecule.

With a prediction tool, based on artiﬁcial intelligence, in [17], we parti-

tioned all the possible hexapeptides into two classes: “amyloidogenic” and “non-

amyloidogenic” with more than 84% correctness. The prediction tool, which

applies a linear Support Vector Machine (SVM) [18], was trained by the Waltz

dataset of 514 amyloidogenic and 901 non-amyloidogenic hexapeptides [19, 20].

The corresponding web-based tool is available as the Budapest Amyloid Predic-

tor at https://pitgroup.org/bap.

As a surprising application of the Budapest Amyloid Predictor, numerous

amyloid- and non-amyloid hexapeptide patterns were identiﬁed in [21]. For ex-

ample, it was shown that for all independent amino acid substitutions for the po-

sitions marked by “x”, the CxFLWx or FxFLFx patterns describe amyloidogenic

hexamers, while the patterns PxDxxx or xxKxEx describe non-amyloidogenic

hexamers. Note that each pattern with two x’s describes 202= 400, while those

with four x’s describe 204= 160,000 diﬀerent hexapeptides succinctly if for the

positions x, we can substitute the 20 amino acid residues.

In the present work, we study the preﬁxes and suﬃxes of the β-sheet sub-

sequences of the PDB-deposited structures and apply the Budapest Amyloid

Predictor to the hexamers of the border regions on these β-sheets. As we show

below, the number of the amyloidogenic hexapeptides of the border regions of the

β-sheets is one-tenth of the same number in the inner parts of those sequences

by screening the homology-ﬁltered Protein Data Bank entries. We believe that

this ﬁnding is a very strong evidence for our hypothesis.

We remark that applying the artiﬁcial intelligence-based Budapest Amyloid

Predictor for the 64 million (= 206) possible hexapeptides is a crucial step

here: the largest experimental dataset that labels the hexapeptides by their

amyloidogenic propensity, the Waltz dataset, contains only 1415 hexapeptides

[19, 20], while we analyzed the preﬁxes and suﬃxes of more than 110,000 β-

sheets from the homology-ﬁltered PDB. Gaining experimental data for that

many hexapeptides is intractable today.

Methods

For the statistical analysis involving the entries of the Protein Data Bank

(PDB) [16], usually some “non-redundant” homology-ﬁltered PDB-subsets are

applied because otherwise, the multiple depositions of more important or more

researched protein structures (e.g., several thousand copies of the HIV-1 pro-

tease) would impact the results. In this study, we have made use of the represen-

tative polypeptide chains of the 30% homology-ﬁltered set, which was available

at the URL https://www.rcsb.org/pdb/rest/representatives?cluster=30

(downloaded on May 21, 2020).

Technical note for reproducibility of the present work: At the time of writing

this section, we were informed that the administrators of the RCSB PDB have

removed the link target indicated above, but one can still generate the similar

non-redundant set with an advanced search feature of the RCSB PDB, namely

https: // search. rcsb. org/ #group-by-context , by choosing 30% sequence

identity.

The structure of almost all representative members of this non-redundant

set was identiﬁed in a soluble state, so almost all proteins in the set are water-

soluble (see Figure S3 in the supporting material).

From the representative sequences of the non-redundant set, we have identi-

ﬁed the maximal contiguous subsequences corresponding to β-sheets by applying

the “SHEET” records in the PDB ﬁle. In other words, all the non-expandable

contiguous β-sheet subsequences were identiﬁed. Those subsequences whose

lengths are less than 6 were discarded. This way, we have identiﬁed as many as

117,467 subsequences.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

OpeningAmyloid-WindowstotheSecondaryStructureofProteins:TheAmyloidogenecityIncreasesTenfoldInsideBeta-SheetsKristofTakacsa,BalintVargaa,ViktorFarkasc,AndrasPerczelc,d,VinceGrolmusza,b,aPITBioinformaticsGroup,EotvosUniversity,H-1117Budapest,HungarybUratimLtd.,H-1118Budapest,HungarycELKH-ELTEPr...

展开>> 收起<<

Opening Amyloid-Windows to the Secondary Structure of Proteins The Amyloidogenecity Increases Tenfold Inside Beta-Sheets.pdf

共12页,预览3页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Opening Amyloid-Windows to the Secondary Structure of Proteins The Amyloidogenecity Increases Tenfold Inside Beta-Sheets

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: