Fator Neurotrófico Derivado Do Cérebro (BDNF)
Fator Neurotrófico Derivado Do Cérebro (BDNF)
Fator Neurotrófico Derivado Do Cérebro (BDNF)
FACULDADE DE BIOCIÊNCIAS
Porto Alegre
2010
EVOLUÇÃO E MODELAGEM MOLECULAR DO
FATOR NEUROTRÓFICO DERIVADO DO CÉREBRO (BDNF) EM
MAMÍFEROS
Porto Alegre
2010
"I believe that the study of science, the learning of the scientific method
by all people, will ultimately help the people of the world in the solution of
our great social and political problems."
Linus Pauling
AGRADECIMENTOS
“Sou parte da história de vocês, na imensa narrativa de suas vidas. Portanto esta
sub-parte Pós-Graduação, também é, concomitantemente, esforço e dedicação de
vocês! Amo vocês.”
Eduardo Eizirik
Osmar Norberto de Souza
i
LISTA DE FIGURAS E TABELA
ii
LISTA DE SIGLAS
BLAST: Basic local alignment search tool (Ferramenta básica de alinhamento local)
iii
RESUMO
ABSTRACT
The neurotrophic factor known as BDNF has attracted considerable attention in the
scientific literature, due to its involvment in critical processes related to the development
and regulation of the nervous system. However, many aspects of this molecule are still
unknown, while others have only been investigated in humans and model organisms. In
this context, bioinformatic tools can be very useful to perform comparative analyses
focusing on evolutionary, structural and functional aspects of this protein. This study
aimed to characterize the patterns of conservation and variability in this molecule
through the comparison of coding sequences in different mammalian lineages, as well
as to investigate structural aspects by constructing a novel 3D model of a BDNF
homodimer that is consistent with experimentally validated data. Our results revealed
very interesting patterns of conservation and variability in different lineages, including
strong evidence for the occurrence of negative natural selection (indicating constraints to
evolutionary change implying functional relevance) in both the mature and the pro
domains. In addition, we observed evidence of positive selection (induction of changes
that are likely adaptive) in different lineages, in every case directly affecting the pro
domain. These results indicate that the pro domain presents great functional relevance,
and highlight the importance of focusing research efforts on this portion of BDNF.
iv
ÍNDICE
Agradecimentos…………………………………………..………...................................i
Lista de Siglas...........................................................................................................iii
Resumo......................................................................................................................iv
1.7. Objetivos........................................................................................................23
Referências...............................................................................................................60
5
Capítulo 1 – Introdução e Objetivos.
rotas funcionais distintas, porém paralelas (Hallböök, 1999). O BDNF foi o segundo fator
trófico a ser identificado e caracterizado (Barde et al., 1982), antecedido pelo NGF; esse
lideraram a formação desta família; após um episódio inicial de duplicação, um dos genes
descendentes deu origem ao par BDNF e NT-4, enquanto o outro ao par NT-3 e NGF
(Hallböök, 1999; Lanave et al., 2007). Só após o segundo evento de duplicação houve a
revela cerca de 40% de similaridade entre os membros das neurotrofinas (Figura 1), e
estão agrupadas em blocos (Robinson et al., 1995; Butte et al., 1998; Robinson et al.,
1999).
1
Figura 1. A Figura representa as quatro sequências protéicas da família das neurotrofinas utilizadas para
os experimentos de raios-X. Legenda: Precursor NGF-beta (241 aa); Precursor BDNF (247 aa); Precursor
NT-3 (257 aa); Precursor NT-4 (210 aa), com suas respectivas identificações nos bancos de dados (gi:
sequence identifiers; an: acession number); * região nas seqüências alinhadas de identidade de
aminoácido; : região nas seqüências alinhadas de similaridade de aminoácidos; . região nas seqüências
alinhadas de aminoácidos não-relacionados. O formato das seqüências utilizado para o alinhamento foi
Pearson, e a matrix blosum.
membros também exercem uma ampla gama de ações rápidas, incluindo efeitos
2
1.2. Estrutura e expressão do gene BDNF.
LIN-7 (GeneID: 55327), que codifica para uma proteína pequena descoberta em
da classe de proteínas heat shock, conhecido como HSP90AA2 (GeneID: 3324); essa
espécies, destacando os mamíferos, foco desse estudo, com mais de 203 sequências
estrutura do gene BDNF, estudada em roedores (Aid, et al., 2007), demonstrou que
camundongo e rato têm oito exons 5´, e um exon 3´, sendo que este último codifica para a
3
foram detectados transcritos no Sistema Nervoso Periférico - assim como associado a um
amplo padrão de expressão, por exemplo, no coração, rim, músculo, próstata, placenta,
roedores.
quase 30 anos (Barde, 1982). O estudo comparou esta proteína purificada com o NGF –
durante o desenvolvimento, mas persiste em muitas partes do cérebro adulto, sendo mais
prováveis rotas para sua sinalização, visto sua grande importância para a pesquisa em
neurociências.
isoforma a perfaz a maioria das pesquisas. Esta isoforma constitui-se em uma pré-
peptídeo sinal de direcionamento) (Figura 2B) possui dois motivos de clivagem; o primeiro
origina uma forma curta do proBDNF (Figura 2C), e o segundo uma forma madura (Figura
2D), ambas independentes (Mowla et al., 2001; Lanave et al., 2007). Em relação à
4
primeira clivagem, a forma longa do pro-BDNF é processada intracelularmente através da
54
proteinase SKI-1, na qual reconhece o motivo consenso RGLT↓SL59 de resíduos não-
57
básicos, e cliva exatamente no aminoácido T, originando a forma curta do proBDNF.
Ellis, 2008), para constituir o peptídeo maduro de 119 aminoácidos. Outro trabalho
motivo, o BDNF é direcionado para uma via constitutiva (Lou et al, 2005).
Figura 2. O BDNF é traduzido em uma pré-proproteína, constituída de um pré-prodomínio, um pro-domínio e uma forma madura C-
terminal. A) Pré-proprodomínio com o peptídeo sinal (18 primeiros aminoácidos) para direcionamento ao retículo endoplasmático,
mostrando as posições relativas às clivagens (57 e 128) e um sítio para glicosilação na posição 121. B) Pro-domínio com
características de uma proteína secretória de 229 aminoácidos de comprimento. C) proBDNF curto com 190 aminoácidos de
comprimento, gerada no retículo endoplasmático. A clivagem pelo pró-hormônio convertase S1P/SKI-1ocorre na posição 57T, no
motivo consenso 54RGLT↓SL59. D) Forma madura com 119 aminoácidos de comprimento. A clivagem pela plasmina ocorre na posição
128
R, no motivo consenso 125RVRR↓128. Esquematizada na figura, está à localização aos pares dos seis resíduos de cisteína que
formam as três pontes dissulfeto.
5
A utilização do proNGF recombinante para elucidar a relação da sua porção pro –
peptídica com o domínio maduro foram importantes a fim de obter uma co-relação com o
físico-químicas revelaram que a região pro do NGF forma um dímero, assim como o
subseqüente, sugeriram que a região pro do NGF parece estruturar-se quando se liga a
parte madura, adotando uma estrutura terciária definida. Considerando o enfoque nas
intramolecular dos domínios pro e maduro do NGF (W -83 - A-63 na região pro; W 121 na
colaboradores (2008), mensurando esforços para obter uma delimitação estrutural para a
na maior parte, regiões desordenadas intrinsecamente (IDRs). Com base nesta inferência,
sua função.
1
SAXS (Small-Angle X-ray Scattering), método estrutural aplicável a partículas nativas em solução; é adequado,
particularmente, ao estudo de sistemas menos estruturados. Metodologia comumente usada para pesquisar processos,
como por exemplo, relacionados ao enovelamento/não-enovelamento de proteínas, ou rotas não totalmente elucidadas
(Bernadó et al., 2007).
6
Em relação ao processamento, dois importantes estudos foram relatados: no
se que a endositose, assim como a clivagem são pré-requisitos para a produção da forma
processada, capaz de induzir a ativação do receptor Trk, e inferiu-se que sob condições
formam três pontes disulfeto (Robinson et al., 1995). Os referidos resíduos de cisteína, na
de dímero, com uma interfase dimérica formada por interações hidrofóbicas conservadas,
formada por sete folhas-β, das quais quatro, formam dois pares de folhas-β antiparalelas
unidos por três loops do tipo grampo (P-hairpin) (loop 1, loop 2, e loop 4), e um loop longo
7
Figura 3. Topologia de um monômero do matBDNF. O esquema
mostra os quatro loops, assim como, os dois pares de folhas- β
antiparalelas (identificados pelas letras A à D). N e C terminais estão
indicados. Os elementos de estrutura secundária típico desta família
são constituídos de sete folhas- β. Fonte: Butte, 2001.
a importância da região pro no direcionamento para uma rota distinta da forma madura,
assim como no auxílio do enovelamento adequado da forma madura. A esta prévia falta
de suporte para as pesquisas com o pro-BDNF intacto, corrobora a atual situação que
aos receptores TrkB (também conhecido como NTRK2) e p75NTR. Os estudos com os
8
complexos de heterodímeros comprovaram sua existência in vitro, mas a atividade
biológica in vivo ainda não está estabelecida (Robinson et al., 1995; Butte et al., 1998;
Robinson et al., 1999). Esses estudos corroboram a idéia que populações neurais
específicas sofram a ação do BDNF complexado com outras neurotrofinas para reforçar
biológica, em neurônios; este estudo concluiu que houve maior atividade se considerado o
BDNF como homodímero (duas unidades simétricas) (Junkbluth et al., 2005). Entretanto,
no contexto estrutural, ainda não existe um modelo cristalográfico que englobe dois
porção pro.
Tabela 1. Estruturas cristalográficas disponíveis com uma porção do gene BDNF, e a porção madura da proteína (matBDNF -
resíduos 8-116).
(Gray & Ellis, 2008). Em relação à estabilidade dimérica do pro-peptídio, há uma hipótese
2
Seqüência do bdnf compreendendo os nucleotídeos 1 a 40:
Região 5´ TCTGGAACGGAATTCTTCTAATAGAAGAATTCCGTTCCAG 3'.
9
que esta região adota uma estrutura terciária estável dimérica, por estabilizar o padrão
Nos últimos anos, houve maior interesse em relação ao fato que o proBDNF e o
co-receptor sortilina, favorece uma rota pró-apoptótica (Teng et al., 2005). A ligação do
o proBDNF secretado sinaliza para uma via constitutiva (Labmann & Brigadski, 2009). A
Figura 4 mostra a interação do BDNF com algumas proteínas, e o tipo de evidência para
10
clivagem do proBDNF é regulada ao longo do desenvolvimento, e sua ação pode ser mais
Figura 4. A figura mostra um esquema da interação do BDNF com outras proteínas. Legenda: TH:
Tirosina 3-Monooxigenase; GDNF: fator neurotrófico derivado da glia; S1P: Precursor da Protease
serínica sítio 1; NTRK1: Precursor do receptor do NGF de alta afinidade (também conhecido como
TRKA); NGFR: Receptor do NGF; NTRK2: precursor do receptor dos fatores BDNF e NT4; SORT1:
Precursor da sortilina (também conhecido como NTR3 – receptor da neurotensina tipo 3); TRKC:
precursor do Receptor do fator NT3; NTF3: Neurotrofina tipo 3; NTF4: Neurotrofina tipo 4 (também
conhecido como NTF5 – Neurotrofina tipo 5). A cor das linhas representa o tipo de evidência para a
associação: Experimentos (em rosa), informações em bancos de dados (em azul), mineração de
dados (verde-musgo) e, homologia (roxo claro). Fonte: Webserver SMART. Disponível em:
www.ebi.ac.uk/embl.
matBDNF é enfatizada por sua contribuição a uma faixa de respostas adaptativas neurais,
incluindo LTP, certas formas de plasticidade de curto prazo, bem como na regulação
2004). Já, o oposto ocorre com o proBDNF, que quando administrado exogenamente,
11
de longo-termo (LTD) no hipocampo (Woo, et al., 2005). Estudos mais recentes e
(Negappan, et al., 2009). Dois SNPs raros, resistentes a clivagem para a forma madura –
Parkinson (Howells et al., 2000) e Epilepsia (Binder et al., 2001). Outros trabalhos
seqüência do BDNF, é uma substituição da valina (Val) pela metionina (Met) no códon 66
constataram que essa variante foi expressa em níveis normais no cérebro, mas sua
al., 2007).
12
1.5. Alinhamento de seqüências e caracterização molecular evolutiva.
seqüências por meio de buscas por uma série de caracteres ou padrões de caracteres
comparação entre diferentes espécies, busca pela correta resolução taxonômica, assim
como busca por motivos seqüenciais que envolvam aspectos funcionais. O alinhamento
gaps (representados pelo símbolo -), através da obtenção de escores para avaliar o
como matriz Blosum (Lesk, 2008). Esta última é o método mais indicado para
pelas referidas seqüências é feita utilizando algoritmos, como por exemplo, BLAST
(Ewens & Grant, 2001). Este método permite comparar a seqüência-alvo com uma
rápido;
13
Uma porção do gene BDNF tem sido usada como marcador filogenético em
diversos estudos de mamíferos, o que levou a um acúmulo de dados sobre estes locos.
filogenia de algumas famílias ou espécies dentro do grande grupo não está bem resolvida.
futuros estudos de genômica comparativa entre as famílias, e para auxiliar futuros estudos
correta classificação entre elas. Juntamente com outras analises utilizando os outros
demonstrado que estas seqüências são bem adequadas para analises filogenéticas dos
mamíferos, podendo resolver a relação basal entre eles (Kullander et al., 1997). Já em
dentre elas o BDNF, a qual tem colaborado para estudos de interpretação do padrão
desta infraclasse (Murphy et al., 2001). Dentro dos mamíferos placentários, os morcegos
(ordem Quiróptera) perfazem mais que 20% dos mamíferos existentes (Miller-Butterworth,
morcegos (Kawai et al. 2002; Van Den Bussche & Hoofer, 2004; Eick et al. 2005). Após
14
um crescente esforço, através da análise das seqüências de representantes das famílias
de morcegos, com base em 16 genes nucleares, entre eles o BDNF, houve a confirmação
sua história filogenética (Givnish & Sytsma,1997). Sendo assim, utilizando-se da análise
classificados dentro de quatro maiores clados e três linhagens monotípicas (Koepfli, et al.,
2008). Através da análise desse quadro de classificação houve a inferência que esta
família sofreu dois períodos de diversificação coincidindo com a maior mudança paleo-
positiva ocorrem taxas de mudanças que seriam maiores que as encontradas ao acaso,
funcional, sendo que o tipo mais comum de ser detectado ao longo do DNA é aquela que
15
códons, que diferem no número de substituições não-sinônimas por códon não sinônimo
sobre o número de substituições sinônimas por códon sinônimo (dN/dS) conhecido como
estimativa de ômega (ω). O modelo M0 assume o mesmo valor de ômega para todos os
outro valor de ω igual a 1, que indica códons neutros; M2 assume ω>1, inferindo a
discreta (M0 e M3), e distribuição beta (M7 e M8). Os modelos aos pares são usados para
calcular o teste de razão de verossimilhança (Likelihood Ratio Test – LRT) e indicar qual
conservado (Nayeem et al., 2006) são utilizadas para modelar proteínas-alvo, na qual se
modelagem comparativa (Martí-Renom et al., 2000; Deane & Blundell, 2003). A seqüência
Bank) (Berman et al., 2000), a fim de escolher a(s) estruturas tridimensionais - template(s)
- mais apropriada(s) aos objetivos de uma pesquisa. Estes três primeiros procedimentos:
16
conduzidos através de uma busca no banco de dados protéico online do National Center
blast para a procura por seqüências homólogas, e a matriz de substituição bluson45 para
BLAST são então utilizados para a obtenção de um alinhamento ótimo, como o programa
exemplo, Modeller, na qual é o mais utilizado para pesquisas em modelagem (Sali &
modelada podem ser detectadas através de campos de força empíricos, e corrigidas por
tendo como base para os cálculos, estruturas tridimensionais resolvidas (Da Silva & Silva,
como por exemplo, comprimento das ligações, ângulos entre as ligações, ângulos
17
torsionais, e quiralidade dos aminoácidos (Höltje, et al., 2003); estes parâmetros podem
confiabilidade dos modelos gerados pode ser avaliada através dos softwares VERIFY3D
(Luethy et al., 1992), Errat (Colovos & Yeates, 1993), e, vizualizados através dos
softwares Swiss-PdbViwer (Guex & Peitsch, 1997; Schwede, et al, 2003), e PyMOL
(DeLano, 2002).
contribuir para elucidar aspectos desconhecidos desta molécula. Neste contexto, análises
maiores detalhes aspectos evolutivos, assim como, redirecionar os estudos com esta
neutra). Esta análise foi relacionada a uma caracterização estrutural paralela para abordar
18
1.7. Objetivos.
evolutivas.
análises estruturais.
19
Capítulo 2 – Artigo Científico
20
Brief Communication
Nature Neuroscience
Draft, 20 April 2010.
1
Laboratório de Biologia Genômica e Molecular, Faculdade de Biociências, PUCRS. Porto
Alegre, RS 90619-900, Brazil.
2
Laboratório de Bioinformática, Faculdade de Informática, PUCRS. Porto Alegre, RS
90619-900, Brazil.
Abstract
Multiple studies have addressed the function of brain-derived neurotrophic factor, yet most
have only included data from humans or model organisms. Here we apply bioinformatics
approaches to model the structure of human proBDNF and to analyze the evolution of this
protein in mammals. We show that strong selective constraint occurs on both the pro and
mature domains of the protein, and find evidence of adaptive evolutionary change in the
former.
Main text
A great many recent studies have focused on the structure and function of the
brain-derived-neurotrophic factor (BDNF), including its implication in human disorders and
analyses of its developmental and regulatory roles in the nervous systems of model
animals1-5. Although this molecule has been clearly implicated in many important functions,
several aspects of its biological roles remain poorly understood. In particular, the precise
role of proBDNF (the longer, pre-cleavage form of the protein) is still contentious6-8, with
limited data available on this molecule relative to the mature form of the peptide
(matBDNF). In addition, very little is still known with respect to the structure and function of
BDNF in non-model species, so that a potential wealth of comparative information on this
important molecule remains virtually untapped.
21
Comparative molecular data may clarify patterns of conservation at the DNA or
protein levels, indicating regions that may be under strong evolutionary constraint (i.e.
negative selection), implying functional relevance, as well as others that bear signatures of
adaptive change (i.e. positive selection)9. In this context, a powerful approach to
understand biological function of target molecules is to combine methods that detect the
strength and pattern of natural selection at each codon with three-dimensional models of
protein structure associated with experimentally validated information on active motifs.
Such a combined approach has not yet been applied to BDNF, despite the interest in
characterizing in detail its structure and function. Moreover, although various structural
aspects of BDNF have been characterized (e.g. refs. 10,11), no current model of its
structure includes the pro domain, hampering assessments of the biological impact of
protein variants affecting this region of the molecule.
In parallel with the efforts to characterize BDNF structure and function, partial or
complete DNA sequences from the gene encoding this protein have been used in many
phylogenetic studies. The first of those studies, published by our group in 2001, addressed
the relationships among placental mammals12, and led to an intriguing observation
regarding the structure of BDNF: two different mammalian lineages (caviomorph rodents
such as guinea pigs, and cetartiodactyls such as cows and pigs) exhibited up to 11 extra
serine residues (relative to the human sequence) inserted at the same position in the
protein (near the cleavage site yielding a shorter proBDNF – see Table S1; Figs. S1, S4),
rendering the resulting peptide potentially very different from those of model organisms.
This finding led us to investigate in detail the levels of sequence variation in BDNF across
mammals, to perform site-by-site analyses of natural selection, and to construct a novel
structural model of a complete proBDNF peptide onto which to map the detected patterns
of constraint and modification.
We specifically aimed to address five questions: (i) how much selective constraint
(i.e. functional relevance) can be detected across proBDNF? (ii) is there any evidence of
positive selection at any site? (iii) what is the 3D structural distribution of amino acid sites
inferred to be under selection? (iv) are the detected serine expansions located near any
functionally relevant site? (v) is there evidence for different evolutionary forces acting upon
BDNF in different mammalian groups?
To address these issues we undertook a two-pronged approach, which led us to
construct a novel 3D model of homodimeric proBDNF, and to perform multiple
comparative analysis of BDNF sequence variation across mammals (see Supplementary
22
Methods). Briefly, our modeling effort used coordinates from two templates: 2ia8 was
selected to model the pro domain, and 1b8m was used for the mature portion (see Fig. S2;
Tables S2-S3). Our models were consistent with SAXS experiments of proNGF13.
Furthermore, we superimposed the validated mature domain onto different models obtaing
in modeling (n=10), and confirmed the flexibility of the pro domain, which is best
represented as a random coil (Fig. 1; Fig.S3).
The evolutionary analysis of natural selection employed BDNF DNA sequences
from 153 different mammalian species (Tables S4-S9), which were sorted into six different
data sets: (i) data set 1 contained all available complete coding sequences for BDNF
(n=20); (ii) data set 2 was more inclusive, incorporating complete or partial sequences
(spanning > 489 bp) from 94 different species; (iii – vi) specific data sets focused on the
orders Carnivora (n= 84), Cetartiodactyla (n= 28), Chiroptera (n= 28), and Rodentia (n=
15), whose current sampling allowed for detailed comparative analyses. Preliminary
analysis of sequence variation indicated that both the mature and pro regions of BDNF
contained a substantial portion of conserved sites at the DNA and protein levels (Table
S10). Even though conservation was more extreme in the mature portion of the protein,
68% of the amino acids in the pro domain were completely identical across all major
mammalian lineages (data set 1), indicating considerable constraint in this domain as well.
We then performed codon-by-codon analyses of natural selection, employing several
maximum likelihoods and Bayesian approaches (see Supplementary Methods for details).
After selecting the model that provided the best fit to the data (Table S11), we analyzed
the ratio of nonsynonymous (amino acid changing) to synonymous (silent) substitutions
(ω= dN/dS) for each codon in each data set. Ratios significantly below 1 indicate negative
selection (functional constraint), while those significantly above 1 indicate positive
selection.
Many codons in both the pro and mature domains were inferred to be under very
strong negative selection (Fig. 1; Tables S12-S13). All the amino acid sites that have been
experimentally implicated in BDNF function (e.g. cleavage sites and the human codon 66,
involved the well-known Val/Met polymorphism2,5) had ratios close to zero, corroborating
their strong constraint and validating our approach. Many other sites showed similar levels
of constraint, including those adjacent to the observed serine expansions. This indicates
that those sites are functionally relevant, and that the insertion of multiple extra residues in
that region may lead to biologically meaningful consequences. This suggests that
experimental analysis of BDNF in mammals bearing those insertions (e.g. guinea pig, cow,
23
pig) may provide novel insights onto the function and regulation of this molecule.
Moreover, the full set of codons observed to be under extreme negative selection may be
used to help design experiments that assay their activity directly. Many of these sites are
located in the pro domain, supporting the view that this portion has important biological
functions.
Another interesting result was the detection of five codons in BDNF that may be
under positive selection (Figs. 1 and S4; Tables 1, S12, S13), implying that they have
undergone adaptive change in at least some of the sampled species. Remarkably, all of
these sites were located in the pro domain, indicating that this portion of the protein has
been influenced by recent adaptive evolution in different mammalian lineages. Site 84
yielded the most consistent pattern, which was detected in most data sets, particularly in
data set two (Table 1), whose sample size provided the most power. Differences among
taxon-specific data sets indicate that some sites may have undergone distinct selective
regimes in different mammalian lineages, possibly due to contrasting evolutionary
pressures affecting the regulation and development of their nervous system. Two
interesting examples are codons 51 and 53 (both located near the cleavage site for a short
proBDNF): the former bears very strong evidence of positive selection in Cetartiodacyla
and negative selection in the other taxa; the latter shows signs of positive selection in
rodents, and negative selection in others. Interestingly both of these orders contain
serine/threonine expansions in that region of the protein (see Tables S1, S12-S13). These
and other results outlined here open up new research avenues allowing further dissection
of the biological roles of BDNF in mammals.
References
1. Lee, R., Kermani, P. Teng, K.K., Hempstead, B.L. Science 294, 1945-1948 (2001).
2. Egan, M.F. et al. Cell 112, 257–269 (2003).
3. Lu, B., Pang, P.T., Woo, N.H. Nature Rev. Neurosci. 6, 603-614 (2005).
4. Hu, Y. & Russek, S. J. Neurochem. 105, 1-17 (2008).
5. Soliman, F. et al. Science. 327, 863-865 (2010).
6. Matsumoto, T. et al. Nature Neurosci. 11, 131-133 (2008).
7. Yang, J. et al. Nature Neurosci. 12, 113-115 (2009).
8. Barker, P.A. Nature Neurosci. 12, 105 – 106 (2009).
9. Yang, Z. Curr. Opin. Genet. Dev. 12, 688 – 694 (2002).
10. Lou, H. et al. Neuron. 45, 245 – 255 (2005).
11. Paiardini, A., Caputa, V. Neuropeptides 42, 205 – 214 (2008).
12. Murphy, W. et al. Nature 409, 614 – 618 (2001).
13. Paolleti, F. et al. Proteins 75, 990 – 1009 (2009).
24
Table 1. Detection of specific amino acid sites in BDNF likely to be under positive selection in at least one of
our mammalian data sets. For each of the five sites in each of the data sets, we indicate the probability of
belonging to a category with ω > 1, as well as the mean posterior estimate of ω, which were calculated with
model M8 and the BEB option in the CODEML program (see Supplementary Methods for details). Site
numbers and the respective amino acids follow the observed states in the human sequence (see Tables S12-
S13 for a complete list of values for all sites).
25
Figure Legend
Figure 1. Three-dimensional structure of a human proBDNF monomer, modeled using pdb templates 1b8m
subunit A and 2ia8. The top portion of each panel corresponds to the mature domain of BDNF, while the
intermediate loop and the bottom portion contain the pro domain. Colors indicate the ω value (i.e. the non-
synonymous to synonymous ratio of nucleotide substitution) for each codon, calculated using data set 1 and
the M8 model (see Supplementary Methods for details). A. Codons exhibiting the lowest ω values (0.053 to
0.06), implying the strongest selective constraint, are shown in black; all others are shown in gray. The arrow
represents the point where serine/threonine expansions have been observed in two mammalian lineages,
exactly where cleavage occurs to form a shorter proBDNF molecule (see Tables S1 and S12). B. Codons
exhibiting the highest values of ω (1.0 to 1.32), implying evidence of positive selection, are shown in black;
all other residues are shown in gray. Arrows indicate three other sites (Asn47, Ala51, and Ser53) bearing ω
values > 1 in other data sets (see Tables 1, S12 and S13).
Figure 1
26
Online Supplementary Information – Mattei et al. 2010
1. Supplementary methods…………………………………………….....2
3. Supplementary Tables……………………….....................................5
4. Supplementary Figures..................................................................34
27
Supplementary Methods
Mammalian BDNF coding sequences (CDS) were obtained from GenBank entries
(http://www.ncbi.nlm.nih.gov), using the BLAST algorithm applied to the nucleotide
collection database. Sequences from 153 species were retrieved, including complete CDS,
partial CDS or full mRNAs. Species names, sequence lengths, GenBank accession
numbers and the original reference are presented in Tables S4-S9. These sequences
were aligned using the CLUSTALW algorithm6, manually inspected, and sorted by
sequence length or taxonomic assignment, representing different mammalian lineages.
Sequences with ambiguous characters were eliminated from the analysis. All analyses
were conducted excluding the signal peptide (first 18 codons). The alignments sorted by
length were divided into two groups: i) data set 1, which included species with complete
coding sequences; and ii) data set 2, which was selected to represent 24 mammalian
lineages (with each sequence spanning at least 489 bp). The alignments sorted by taxon
28
represented mammalian orders which currently present a reasonable sample of available
species: Carnivora, Cetartiodactyla, Chiroptera and Rodentia. Data sets were organized
using the package MEGA 4.07, which was also employed to run exploratory analyses of
sequence conservation.
More detailed molecular evolutionary studies were conducted with the CODEML program
within the PAML4 package8, which provides log-likelihood (lnL) values for different models
of codon evolution and posterior probabilities for distinct selection regimes (negative,
neutral, positive) at each site. Specifically, we provided the program with unrooted
Neighbor-Joining trees estimated in MEGA with Maximum Composite Likelihood
distances, including all codon positions. All positions containing alignment gaps and
missing data were eliminated in a pairwise fashion (pairwise deletion option).
We conducted Likelihood Ratio Tests (LRT) to assess model fitting. These tests employ as
focal statistics twice the difference between the log likelihoods (lnL) of contrasted models:
M1 vs. M2, and M7 vs. M8 (only nested models should be compared with this approach),
with a χ2 null distribution and degrees of freedom equal to the difference in the number of
parameters9. The M7-M8 comparison is a stringent test of positive selection, compared
with M1-M2. The M0-M3 comparison, however, is more a test of variable selective
pressure among sites than a test of positive selection10.
The omega (ω) parameter was estimated using the option NSsites. This specifies models
that allow the dN/dS ratio (ω) to vary among sites11,12. We analyzed six NSsites models
which differed with respect to proportion of site classes (p) and their respective mean ω
value. Moldels were as follows: MO (one-ratio) assumes the same ω ratio for all sites; M1
(Nearly Neutral) assumes ω=0 (conserved sites) and ω=1 (neutral sites); M2 (Positive
selection) allows conserved sites, neutral sites and positive sites, the latter with ω>1
(independent); M3 (Discrete) assumes these same site classes, but following a discrete
distribution, with both the proportion of classes and ω parameters estimated from the data;
M7 (beta) and M8 (beta&ω>1) assume a beta distribution of ω values, with M8 allowing
the presence of positively selected sites. An empirical Bayes (EB) approach was then
used to calculate the posterior probability of each codon belonging to a specific site class.
29
References for the supplementary methods
1. Martí-Renom, M., Stuart, A., Fiser, A. et al.. Annu. Rev. Biophys. Biomol. Struct, 29, 291 – 325 (2000).
2. Guex, N., Peitsch, M. Electrophoresis, 18, 2714 – 23, 1997.
3. DeLano, W.L. The PyMOL Molecular Graphics System (2002) DeLano Scientific, San Carlos, CA, USA.
4. Sali, A., Blundell, T. J. Mol. Biol. 234, 779 – 815 (1993).
5. Laskowski R A, MacArthur M.W, Moss D. S, Thornton J. M. J. App. Cryst. 26, 283 – 291 (1993).
6. Thompson, J., Higgins, D., Gibson, T. Nucleic Acids Research. 22, 4673 – 80 (1994).
7. Tamura K.; Dudley J.; Nei, M; Kumar, S. Mol Biol Evol. 24, 1596 – 99 (2007).
8. Yang, Z. Mol Biol Evol. 24, 1586 -91 (2007).
9. Yang, Z. Mol. Biol. Evol. 15, 568 – 73 (1998).
10. Anisimova, M.; Bielawski, J.; Yang, Z. Mol Biol Evol. 18, 1585 – 92 (2001).
11. Nielsen, R.; Yang, Z. Genetics. 148, 929 – 936 (1998).
12.Yang, Z., Nielsen, R, Goldman, N, Pedersen, A. Genetics. 155, 431 – 449 (2000).
30
Supplementary Tables.
31
Table S2. Number of residues changed and total
number of residues. Information of one proBDNF
monomer (229 residues), extracted from the program
Procheck 3.3.
Parameter Results
Average value of CA – N –C - CB 34.52 +- 1.32
Distorted main-chain bonds 2
Distorted main-chain angles 1
Distorted planar group 0
Neighbouring contacts 771
Number of hidrogen bonds 126
Potential contact atoms 1803
unusual CA-CA distance for contact number 389
distance 16.1
Phi-psi and chi1-chi2 distributions
Ramachandran Plot Statistics (for 200 residues)
[A,B,L]1 89.0%
[a,b,l,p]2 9.5%
[~a,~b,~l,~p]3 1.5%
[XX]4 0.0%
Number of end-residues (excl. Gly and Pro) 2
Number of glycine residues 17
Number of proline residues 10
1 3
Residues in most favored regions Residues in generously allowed regions
2 4
Residues in additional allowed regions Residues in disallowed region
32
Table S4. Sequences of Data Set 1 analyzed in this study.
33
Table S5. Additional sequences included in Data Set 2 (analyzed jointly with those listed on Table S4).
34
Table S5. Continued.
35
Table S5. Continued.
36
Table S6. Sequences of the order Carnivora analyzed in this study.
37
Table S6. Continued.
38
Table S6. Continued.
39
Table S7. Sequences of the order Cetartiodactyla analyzed in this study.
40
Table S8. Sequences of the order Chiroptera analyzed in this study.
41
Table S9. Sequences of the order Rodentia analyzed in this study.
42
Supplementary references for tables S4-S9.
Arab, S.; Krohn, K.; Lachmund, A., et al. The gene encoding bovine brain derived neurotrophic factor
(BDNF), Gene, v.185, pp.95-8, 1997.
Demere, T.; McGowen, M.; Berta, A.; Gatesy, J. Morphological and molecular evidence for a stepwise
evolutionary transition from teeth to baleen in mysticete whales. Syst. Biol., v. 57 n.1, pp. 15-37, 2008.
Dorus, S.; Vallender, E.; Evans, P., et al. Accelerated evolution of nervous system genes in the origin of
Homo sapiens. Cell, v.119, n. 7, pp.1027-40, 2004.
Janecka, J. ; Miller, W.; Pringle, T. et al. Molecular and genomic data identify the closest living relative of
primates. Science, v. 318, v. 5851, pp. 792-94, 2007.
Koepfli, K.; Gompper, M.; Eizirik, E., et al. Phylogeny of the Procyonidae (Mammalia: Carnivora):
Molecules, morphology and the Great American Interchange. Mol. Phylogenet. Evol., v 43 , n.3,
pp.1076-95, 2007.
Koepfli, K.; Deere, K.; Slater, G. et al. Multigene phylogeny of the Mustelidae: resolving relationships,
tempo and biogeographic history of a mammalian adaptive radiation. BMC Biol., v. 6, v.1, pp. 10, 2008.
Leibrock, J.; Lottspeich, F.; Hohn, A., et al. Molecular cloning and expression of brain derived neurotrophic
factor, Nature., v.341, n.6238, pp.149-52, 1989.
Maisonpierre, P.; Belluscio, L.; Friedman, B., et al. NT-3, BDNF, and NGF in the developing rat nervous
system: parallel as well as reciprocal patterns of expression. Neuron, v. 5, n. 4, pp. 501-09, 1990.
Miller-Butterworth, C.; Murphy, W.; O'Brien, S., et al. A Family Matter: Conclusive Resolution of the
Taxonomic Position of the Long-fingered Bats, Miniopterus. Mol. Biol. Evol, v. 24, n.7, pp.1553-61,
2007.
Murphy, W.J.; Eizirik, E.; O'Brien, S.J, et al. Resolution of the early placental mammal radiation using
Bayesian phylogenetics, Science, v.294 , n.5550, pp. 2348-2351, 2001a.
Murphy, W.; Eizirik, E.; Johnson, W., et al. Molecular phylogenetics and the origins of placental mammals,
Nature, v.409, n. 6820, pp. 614-8, 2001b.
Pollinger, J.; Bardeleben, C.; Koepfli, K., et al. Genome Sequence, Comparative Analysis and Haplotype
Structure of the Domestic Dog, Nature, In press, 2005.
Pruunsild, P.; Kazantseva, A.; Aid, T., et al. Dissecting the human BDNF locus: bidirectional transcription,
complex splicing, and multiple promoters. Genomics, v. 90, n.3, pp.397-406, 2007.
Roca, A.; Bar-Gal, G.; Eizirik, E., et al. Mesozoic origin for West Indian insectivores. Nature, v. 429, n.
6992, pp. 649-51, 2004.
Strausberg, R.; Feingold, E.; Grouse, L. et al. Generation and initial analysis of more than 15,000 full-
length human and mouse cDNA sequences. Proc Natl Acad Sci USA, v.99, n. 26, pp.16899-90, 2002.
Teeling, E.; Springer, M.; Madsen, O., et al. A molecular phylogeny for bats illuminates biogeography and
the fossil record. Science, v. 307, n. 5709, pp. 580-84, 2005.
Vischer, H. BDNF is expressed at the crush site after spinal cord lesion in newborn opossum (Monodelphis
domestica). Eur. J. Neurosci., v.9 , n. 9, pp.1993-7,1997.
43
Table S10. Proportion of conserved sites in different BDNF data sets.
44
Table S11. Parameter Estimates, Log-likelihood Values, and Likelihood Ratio Test for the models M1, M2, M7 and M8. More
information about the hypotheses tested can be found on Supplementary Methods. P: number of site classes contained in each
model, including the ω distribution; lnL: log likelihood values; LRT: Likelihood Ratio Test (with α =0.05 and df=1).
Data set 1.
Data set 2.
45
Table S11. Continued.
46
Table S11. Continued.
47
Table S12. Analysis of natural selection in the BDNF coding sequence (sites 1-18
correspond to the signal peptide and were excluded), based on data sets 1 and 2
(see Supplementary Methods). For each residue, we indicate the posterior
probability of belonging to the site class exhibiting evidence of positive selection,
as well as the mean + standard error (SE) of the nonsynonymous to synonymous
substitution ratio (ω ). Sites indicating very low probability of positive selection (i.e.
close to zero) indicate high probability of negative selection (i.e. functional
constraint). Residues are numbered according to the human sequence. Sites
inferred to be under positive selection (see Table 1) are marked with an asterisk.
Sites experimentally demonstrated to be involved in specific aspects of BDNF
function are highlighted in bold, and detailed in the footnotes.
48
Table S12. Continued.
49
Table S12. Continued.
50
Table S12. Continued.
Residues Data set 1 Data set 2
Probab. ω + SE Probab. ω + SE
6
154K 0.00000 0.053 + 0.022 0.00000 0.050 + 0.006
6
155T 0.00004 0.058 + 0.043 0.00000 0.053 + 0.020
6
156A 0.00006 0.059 + 0.046 0.00000 0.052 + 0.017
6
157V 0.00000 0.054 + 0.027 0.00000 0.051 + 0.008
6
158D 0.00000 0.053 + 0.021 0.00000 0.050 + 0.005
6
159M 0.00000 0.053 + 0.020 0.00000 0.050 + 0.006
6
160S 0.00003 0.059 + 0.043 0.00000 0.052 + 0.018
6
161G 0.00000 0.055 + 0.029 0.00000 0.051 + 0.009
6
162G 0.00000 0.054 + 0.027 0.00000 0.051 + 0.007
6
163T 0.00000 0.055 + 0.029 0.00000 0.051 + 0.010
164V 0.00000 0.055 + 0.029 0.00000 0.051 + 0.009
165T 0.00001 0.056 + 0.034 0.00000 0.051 + 0.013
166V 0.00000 0.055 + 0.030 0.00000 0.051 + 0.009
167L 0.00009 0.060 + 0.049 0.00000 0.053 + 0.020
168E 0.00000 0.056 + 0.032 0.00000 0.051 + 0.010
169K 0.00000 0.055 + 0.031 0.00000 0.051 + 0.010
170V 0.00000 0.055 + 0.030 0.00000 0.051 + 0.009
171P 0.00006 0.059 + 0.047 0.00000 0.053 + 0.020
172V 0.00001 0.056 + 0.034 0.00000 0.051 + 0.013
6
173S 0.00116 0.094 + 0.116 0.00000 0.057 + 0.031
6
174K 0.00001 0.056 + 0.035 0.00000 0.051 + 0.013
6
175G 0.00000 0.054 + 0.027 0.00000 0.051 + 0.008
6
176K 0.00009 0.060 + 0.050 0.00000 0.052 + 0.016
6
177L 0.00004 0.059 + 0.044 0.00000 0.052 + 0.015
178K 0.00000 0.053 + 0.022 0.00000 0.050 + 0.006
179Q 0.00003 0.058 + 0.043 0.00000 0.052 + 0.014
180Y 0.00001 0.057 + 0.036 0.00000 0.051 + 0.012
181F 0.00000 0.056 + 0.035 0.00000 0.054 + 0.024
182Y 0.00001 0.056 + 0.035 0.00000 0.054 + 0.022
183E 0.00000 0.053 + 0.022 0.00000 0.050 + 0.005
184T 0.00000 0.055 + 0.030 0.00000 0.051 + 0.010
185k 0.00000 0.053 + 0.022 0.00000 0.051 + 0.011
186C 0.00001 0.057 + 0.037 0.00000 0.052 + 0.015
187N 0.00000 0.055 + 0.031 0.00000 0.051 + 0.012
188P 0.00001 0.058 + 0.040 0.00000 0.051 + 0.014
189M 0.00000 0.053 + 0.020 0.00000 0.053 + 0.019
190G 0.00002 0.058 + 0.041 0.00000 0.056 + 0.029
191Y 0.00001 0.057 + 0.038 0.00000 0.054 + 0.023
192T 0.00003 0.058 + 0.042 0.00000 0.052 + 0.015
193K 0.00000 0.054 + 0.025 0.00000 0.051 + 0.008
194E 0.00000 0.054 + 0.024 0.00000 0.050 + 0.007
195G 0.00000 0.055 + 0.030 0.00000 0.051 + 0.009
196C 0.00001 0.057 + 0.036 0.00000 0.052 + 0.016
197R 0.00000 0.054 + 0.027 0.00000 0.051 + 0.011
198G 0.00000 0.054 + 0.027 -
199I 0.00000 0.056 + 0.033 -
200D 0.00000 0.053 + 0.021 -
201K 0.00000 0.055 + 0.029 -
-
202R 0.00000 0.054 + 0.024
51
Table S12. Continued.
53
Table S13. Continued.
Residues Carnivora Cetartiodactyla Chiroptera Rodentia
Probab. ω + SE Probab. ω + SE Probab. ω + SE Probab. ω + SE
45S 0.00076 0.068 + 0.080 0.00068 0.068 + 0.081 0.00001 0.057 + 0.037 0.00001 0.057 + 0.037
46V 0.10198 0.438 + 0.434 0.03558 0.216 + 0.353 0.07736 0.530 + 0.339 0.00337 0.177 + 0.185
47N * 0.24152 0.662 + 0.578 0.00427 0.082 + 0.144 0.61496 1.169 + 0.464 0.02816 0.274 + 0.296
48G 0.00038 0.065 + 0.069 0.00056 0.067 + 0.077 0.00000 0.056 + 0.033 0.00000 0.057 + 0.036
49P 0.00250 0.075 + 0.110 0.00287 0.078 + 0.123 0.00005 0.060 + 0.047 0.00003 0.059 + 0.045
50K 0.00023 0.064 + 0.063 0.00044 0.066 + 0.073 0.00002 0.071 + 0.067 0.02699 0.368 + 0.280
51A * 0.00613 0.086 + 0.150 0.95891 2.423 + 1.195 0.07566 0.370 + 0.391 0.00476 0.117 + 0.164
52G 0.02089 0.172 + 0.262 0.00827 0.093 + 0.196 0.00033 0.064 + 0.064 0.00187 0.102 + 0.132
53S * 0.00289 0.077 + 0.115 0.01884 0.121 + 0.311 0.37821 0.871 + 0.535 0.84950 1.636 + 0.736
1
54R 0.00719 0.088 + 0.160 0.00773 0.092 + 0.189 0.00043 0.065 + 0.068 0.00043 0.065 + 0.068
1
55G 0.01670 0.162 + 0.243 0.00072 0.068 + 0.082 0.00000 0.057 + 0.036 0.00026 0.084 + 0.094
1
56L 0.00302 0.077 + 0.117 0.00451 0.083 + 0.148 0.00007 0.060 + 0.050 0.00281 0.109 + 0.146
1
57T 0.00410 0.080 + 0.130 0.00110 0.071 + 0.091 0.00004 0.059 + 0.045 0.01467 0.233 + 0.250
1,2
58S 0.00301 0.077 + 0.117 0.00381 0.081 + 0.138 0.02350 0.169 + 0.264 0.23282 0.637 + 0.589
1
59L 0.00485 0.082 + 0.138 0.00685 0.089 + 0.180 0.00028 0.063 + 0.061 0.00014 0.062 + 0.056
60A 0.00351 0.079 + 0.123 0.00567 0.086 + 0.165 0.00037 0.064 + 0.065 0.00070 0.067 + 0.076
61D 0.00027 0.064 + 0.065 0.00043 0.066 + 0.072 0.00000 0.054 + 0.027 0.00000 0.055 + 0.029
62T 0.00481 0.082 + 0.137 0.00871 0.095 + 0.201 0.00081 0.067 + 0.078 0.00063 0.066 + 0.074
63F 0.00597 0.085 + 0.149 0.18709 0.572 + 0.871 0.00054 0.065 + 0.071 0.00013 0.062 + 0.055
64E 0.00312 0.077 + 0.118 0.00374 0.080 + 0.137 0.00002 0.059 + 0.043 0.00003 0.059 + 0.045
65H 0.00104 0.070 + 0.086 0.00198 0.074 + 0.109 0.00001 0.057 + 0.038 0.00001 0.057 + 0.038
3
66V 0.00047 0.066 + 0.072 0.00081 0.069 + 0.084 0.00000 0.056 + 0.033 0.00001 0.057 + 0.036
67I 0.00245 0.075 + 0.109 0.00084 0.069 + 0.085 0.00003 0.059 + 0.044 0.00001 0.057 + 0.037
68E 0.00312 0.077 + 0.118 0.00248 0.076 + 0.118 0.00003 0.059 + 0.045 0.00002 0.058 + 0.041
69E 0.00019 0.063 + 0.061 0.00043 0.066 + 0.072 0.00001 0.068 + 0.060 0.00000 0.055 + 0.030
70L 0.00302 0.077 + 0.117 0.00393 0.081 + 0.139 0.00007 0.060 + 0.050 0.00015 0.062 + 0.056
54
Table S13. Continued.
55
Table S13. Continued.
56
Table S13. Continued.
57
Table S13. Continued.
Residues Carnivora Cetartiodactyla Chiroptera Rodentia
Probab. ω + SE Probab. ω + SE Probab. ω + SE Probab. ω + SE
148V 0.00046 0.066 + 0.072 0.00081 0.069 + 0.084 0.00005 0.060 + 0.048 0.00002 0.058 + 0.040
149T 0.00805 0.090 + 0.168 0.00098 0.070 + 0.088 0.00015 0.062 + 0.056 0.00006 0.060 + 0.049
150A 0.00049 0.066 + 0.073 0.00106 0.070 + 0.090 0.00001 0.058 + 0.041 0.00001 0.057 + 0.039
151A 0.00111 0.070 + 0.088 0.00088 0.069 + 0.086 0.00006 0.060 + 0.048 0.00051 0.065 + 0.070
152D 0.00120 0.070 + 0.089 0.00399 0.081 + 0.140 0.00002 0.058 + 0.041 0.00005 0.060 + 0.047
153K 0.00397 0.080 + 0.128 0.00423 0.082 + 0.145 0.00005 0.060 + 0.048 0.00031 0.064 + 0.065
6
154K 0.00023 0.064 + 0.063 0.00044 0.066 + 0.073 0.00000 0.055 + 0.029 0.00000 0.056 + 0.033
6
155T 0.00481 0.082 + 0.137 0.00893 0.095 + 0.204 0.00066 0.066 + 0.074 0.00050 0.065 + 0.070
6
156A 0.00829 0.091 + 0.170 0.00747 0.091 + 0.188 0.00025 0.063 + 0.060 0.00060 0.066 + 0.073
6
157V 0.00046 0.066 + 0.072 0.00081 0.069 + 0.084 0.00000 0.056 + 0.033 0.00001 0.057 + 0.036
6
158D 0.00027 0.064 + 0.065 0.00043 0.066 + 0.072 0.00000 0.054 + 0.027 0.00000 0.055 + 0.029
6
159M 0.00021 0.064 + 0.062 0.00043 0.066 + 0.073 0.00000 0.055 + 0.028 0.00000 0.056 + 0.032
6
160S 0.00296 0.077 + 0.116 0.00594 0.087 + 0.167 0.00013 0.062 + 0.055 0.00019 0.063 + 0.059
6
161G 0.00073 0.068 + 0.079 0.00087 0.069 + 0.085 0.00000 0.057 + 0.036 0.00013 0.062 + 0.055
6
162G 0.00038 0.065 + 0.069 0.00059 0.067 + 0.078 0.00000 0.056 + 0.033 0.00001 0.057 + 0.038
6
163T 0.00063 0.067 + 0.077 0.00206 0.075 + 0.110 0.00000 0.057 + 0.037 0.00007 0.060 + 0.050
164V 0.00086 0.069 + 0.082 0.00110 0.070 + 0.091 0.00001 0.057 + 0.037 0.00001 0.057 + 0.036
165T 0.00091 0.069 + 0.083 0.00122 0.071 + 0.094 0.00001 0.057 + 0.037 0.00121 0.069 + 0.087
166V 0.00086 0.069 + 0.082 0.00102 0.070 + 0.089 0.00001 0.057 + 0.037 0.00001 0.057 + 0.037
167L 0.00671 0.087 + 0.156 0.00405 0.082 + 0.141 0.00058 0.066 + 0.072 0.00013 0.062 + 0.055
168E 0.00312 0.077 + 0.118 0.00465 0.083 + 0.150 0.00003 0.059 + 0.044 0.00003 0.059 + 0.043
169K 0.00385 0.080 + 0.127 0.00491 0.084 + 0.153 0.00003 0.059 + 0.045 0.00002 0.058 + 0.042
170V 0.00086 0.069 + 0.082 0.00102 0.070 + 0.089 0.00001 0.057 + 0.037 0.00001 0.057 + 0.036
171P 0.00401 0.080 + 0.129 0.00291 0.078 + 0.124 0.00096 0.068 + 0.081 0.00007 0.060 + 0.050
172V 0.00611 0.086 + 0.150 0.00102 0.070 + 0.089 0.00009 0.061 + 0.051 0.00008 0.060 + 0.050
6
173S 0.00355 0.080 + 0.134 0.00011 0.061 + 0.053 0.00011 0.061 + 0.054
58
Table S13. Continued.
59
Supplementary Figures.
Fig. S1. Human BDNF isoform a, that codes for a protein containing 247 amino acids. The
first 18 residues comprise the signal peptide. The underlined motifs represent sites within
BDNF that have been implicated in functional aspects: cleavage to form a shorter
proBDNF (54RGLT↓SL59) (Seidah et al., 1999); mature form cleavage (125RVRR128)
(Lee et al., 2001; Gray & Ellis, 2008); carboxipeptidase-E co-receptor interaction (144I and
146E; 233ID234) (Luo et al, 2005); TrkB receptor interaction (154KTAVDMSGGT163;
173SKGQL177; 207QCRTTQSYVR216) (Ibáñez et al, 1991; Ibáñez et al, 1993);
interaction with receptors p75 and TrkB (223KKR225) (Rydén et al., 1995).
60
Figure S2. Alignment of the target-sequence (human locus BDNF - p23560) - excluding
the signal peptide - and the two templates, which have 3D structures available in the
protein data bank (pdb): 1b8m subunit A and 2ia8. 2ia8 subunit A, derived from
mitochondrial cytochrome c peroxidase, was identified as the best available match using
BLAST screening, with 30% identities. 1b8m subunit A was obtained directly from BDNF
crystallographic studies (for more information see Supplementary Methods). The figure
was generated using the multiple sequence alignment program T-Coffee. Symbols: “*”
indicates regions of amino acid identities; “:” indicates regions of amino acid similarities; “.”
indicates regions of unrelated amino acids.
61
Figure S3. ProBDNF homodimer, with two symetric units of the mature domain
(green and red colors), and two symmetric units of the pro domain (blue and yellow
colors). The figures were generated using the program Pymol (version 0.99rc6). A
and B represent distinct views.
62
Fig. S4. Location of positively selected amino acids (underlined) identified in each data
set, using one representative sequence (without the signal peptide) as an example. Grey
fonts indicate the pro domain, whereas black fonts depict the mature domain. A. Protein
sequence of Homo sapiens (data set 1); B. Protein sequence of Homo sapiens (data set
2); C. Protein sequence of cattle (Bos taurus, order Cetartiodactyla); the rectangle
indicates an insertion of three serines in this sequence; D. Protein sequence of a mountain
paca (Agouti taczanowskii, order Rodentia). The rectangle indicates an insertion of 11
serines observed in this sequence; E Protein sequence of sheath-tailed bat (Emballonura
atrata, order Chiroptera); F. Protein sequence of a domestic cat (Felis catus) (order
Carnivora).
63
Capítulo 3 – Considerações Finais.
estáveis com outros membros da família (BDNT/NT3; BDNF/NT4), ainda não havia
Recentes estudos de Paoletti et al. (2006) e Paoletti et al. (2008) demonstraram que
o domínio pró não processado do NGF (membro mais conhecido desta família
protéica) forma também uma estrutura dimérica simétrica. Estes estudos apontaram
64
com o formato crab-like, descrito por estes autores, na qual um monômero é
situados no domínio maduro. Através das nossas análises, tanto o domínio pró
uma maior atenção funcional e estrutural para o domínio pro, do qual se conhece
pouco, como também sua relação com o domínio maduro. Neste contexto, a função
para a proteína madura, e restrições práticas para o estudo in vitro utilizando o pró-
domínio, têm dificultado o direcionamento do foco das pesquisas para a região pro.
substituição única do aminoácido valina por uma metionina no domínio pró. Além de
processado para uma rota de sinalização regulada. Outro intrigante fato descrito diz
65
respeito à inserção dos aminoácidos serina e/ou treonina na seqüência pró de
variação de inserção de uma até onze serinas. Em roedores utilizados como modelo
para os estudos acerca do BDNF (Mus musculus, Rattus norvegicus) foi observado
à inserção de uma treonina seguida pela serina. Através do parâmetro de ômega (ω)
indica que a presença destes aminoácidos adicionais pode ter efeitos biológicos
66
Referências Bibliográficas.
Aid, T.; Kazantseva, A.; Piirsoo, M. et al. Mouse and rat BDNF gene structure and expression
revisited. J. Neurosci. Res., v. 85, pp. 525–35, 2007.
Ammirante, M.; Rosati, A.; Gentilella, A. et al. The activity of hsp90 alpha promoter is regulated by NF-
kappa B transcription factors. Nature, v. 27, n.8, pp. 1175-8, 2007.
Arab, S.; Krohn, K.; Lachmund, A. et al. The gene encoding bovine brain-derived neurotrophic factor
(BDNF). Gene, v.185, n.1, pp. 95-8, 1997.
Balasubramanian, S.; Zheng, D.; Liu, Y. et al. Comparative analysis of processed ribosomal protein
pseudogenes in four mammalian genomes, Genome Biol., v.10, n.1, pp. R2, 2009.
Barde, Y.A; Edgar, D; Thoenen, H. Purification of a new neurotrophic factor from mammalian brain.
EMBO J., v.1, n.5, pp 549-53, 1982.
Berman, H.; Westbrook, J.; Feng, Z. et al. The Protein Data Bank. Nucl. Ac. Res., v.28, pp. 235- 242,
2000.
Bernadó, P.; Mylonas, E.; Petoukhov, M. et al. Structural characterization of flexible proteins using
small-angle X-ray scattering. J Am Chem Soc, v.129, n. 17, pp. 5656-64, 2007.
Binder, D.; Croll, S.; Gall, C. et al. BDNF and epilepsy: too much of a good thing?. Trends in
Neurosci., v. 24, pp. 47-53, 2001.
Bolamos, C.; Nestler, E. Neurotrophic mechanisms in drug addiction. NeuroMol. Med., v. 5, pp. 69-83,
2004.
Boutilier, J.; Ceni, C.; Pagdala, P. et al. Proneurotrophins require endocytosis and intracellular
proteolysis to induce TrkA activation. JBC, v. 263, n. 19, pp. 12709 – 16, 2008.
Butte, M.; Hwang, P.; Mobley, W.; Fletterick, R. Crystal structure of neurotrophin-3 homodimer shows
distinct regions are used to bind its receptors. Biochem, v. 37, pp. 16846-52, 1998.
Butte, M. Neurotrophic factor structures reveal clues to evolution, binding, specificity, and receptor
activation. Cell Mol Life Sci, v. 58, pp. 1003-13, 2001.
Chen, Z.; Jing, D.; Bath, K. et al. Genetic Variant BDNF (Val66Met) Polymorphism Alters Anxiety-
Related Behavior, Science, v. 314, pp. 140 – 43, 2006.
Colovos, C; Yeates, TO. Verification of protein structures: patterns of nonbonded atomic interactions.
Protein Sci., v. 2, n. 9, pp.1511-9, 1993.
Da Silva, V; Silva, C. Modelagem molecular de proteínas-alvo por homologia estrutural. REF, v.4, n.1,
PP. 15-26, 2007.
Deane, C.; Blundell, T. Protein comparative modeling and drug discovery. In: WERMUTH, C. The
practice of medical chemistry. London: Elsevier Academic Press, 2003.
DeLano, W. The PyMOL Molecular Graphics System. DeLano Scientific, San Carlos, CA, USA, 2002.
[http://www.pymol.org].
Egan, M.; Kojima, M.; Callicott, J. et al. The BDNF val66met Polymorphism Affects Activity-Dependent
Secretion of BDNF and Human Memory and Hippocampal Function. Cell, v. 112, pp.257–69, 2003.
Eick, G..; Jacobs, D.; Matthee, C. A nuclear DNA phylogenetic perspective on the evolution of
echolocation and historical biogeography of extant bats (Chiroptera). Mol Biol Evol., v. 22, pp.
1869–1886, 2005.
67
Ferrer, I.; Marín, C.; Rey, M. Et al. BDNF and full-length and truncated TrkB expression in Alzheimer
disease. Implications in therapeutic strategies. J. Neuropathol Exp Neurol., v.58, p.729-39, 1999.
Ewens, W; Grant, G. BLAST. In: Statistical methods in bioinformatics: An introduction (Statistic for
biology and health). Hardcover: Springer, 1ed, 2001.
Givnish T.; Sytsma, K. Molecular Evolution and Adaptive Radiation. Cambridge: Cambridge University
Press; 1997.
Gray, K.; Ellis, V. Activation of pro-BDNF by the pericellular serine protease plasmin. FEBS Lett,
v.582, pp. 907–10, 2008.
Guex, N.; Peitsch, M. SWISS-MODEL and The Swiss-PdbViewer: An environment for comparative
protein modeling. Electrophoresis, v. 18, pp. 2714 – 23, 1997.
Hallböök, F. Evolution of the vertebrate neurotrophin and Trk receptor gene families. Science, v. 9, pp.
616-21, 1999.
Hallböök, F.; Lundin, L.; Kullander, K. Lampetra fluviatilis Neurotrophin Homolog, Descendant of a
Neurotrophin Ancestor, Discloses the Early Molecular Evolution of Neurotrophins in the Vertebrate
Subphylum. J. Neurosci., v.18, n.21, pp. 8700-11, 1998.
Höltje, H.; Sippl, W.; Rognan, D.; Folkers, G. Introduction to comparative protein modeling. In:
Molecular modeling: Basic principles and applications. Weinheim: Wiley-VCH, 2003.
Hong, C.; Huo, S.; Yen, F. et al. Association study of a brain-derived neurotrophic-factor genetic
polymorphism and mood disorders, age of onset and suicidal behavior. Neuropsychobiol., v.48,
pp.186-89, 2003.
Howells, D.; Porritt, M.; Wong, J. et al. Reduced BDNF mRNA Expression in the Parkinson's Disease
Substantia Nigra. Exp. Neuro., v.166, pp. 127-35, 2000.
Jiang, X.; Xu, K.; Hoberman, J. et al. BDNF variation and mood disorders: a novel functional promoter
polymorphism and Val66Met are associated with anxiety but have opposing effects.
Neuropsychopharmacolog., v.30, pp.1353-61, 2005.
Junkbluth, S; Bailey, K; Barde, Y-A. Purification and caracterisation of brain-derived neurotrophic factor
/ neurotrophin-3 (BDNF / NT-3) heterodimer. Eur J Biochem, v. 221, n. 2, pp. 677 – 85, 2005.
Katoh-Semba, R; Takeuchi, I.; Semba, R.; Kato, K. Distribution of brain-derived neurotrophic factor in
rats and its changes with development in the brain. J. Neurochem., v. 69, pp. 34–42, 1997.
Kawai K.; Nikaido M.; Harada M. et al. Intra and interfamily relationships of Vespertilionidae inferred
by various molecular markers including SINE insertion data. J Mol Evol. , v. 55, pp. 284–301. 2002.
Kliemannel, M.; Golbik, R.; Rudolph, R. et al. The pro-peptide of pro-NGF: Structure formation and
intramolecular association with NGF. Protein Science, v. 16, pp. 411 – 19, 2007.
Kliemannel, M.; Rathenholl, A.; Golbik. et al. The mature part of proNGF induces the structure of its
pro-peptide. FEBS Letters, v. 566, pp. 207 – 12, 2004.
Koepfli, K.;Deere, K.; Slater, G. et al. Multigene phylogeny of the Mustelidae: resolving relationships,
tempo and biogeographic history of a mammalian adaptive radiation. BMC Biol., v. 14, pp. 6-10,
2008.
Korneev, S.; O'Shea, M. Natural antisense RNAs in the nervous system. Rev. Neurosci, v. 16, pp.
213-22, 2005.
Koshimizu, H.; Kiyosue, K.; Hara, T., et al. Multiple functions of precursor BDNF to CNS neurons:
negative regulation of neurite growth, spine formation and cell survival. Mol brai, pp. 2-27, 2009.
68
Kovalchuk, Y; Hanse, E.; Kafitz, K.; Konnerth, A. Postsynaptic Induction of BDNF-mediated Long-
Term Potentiation. Science, v.295, pp 1729-34, 2002.
Kullander, K.; Carlson, B.; Hallböök, F. Molecular phylogeny and evolution of the neurotrophins from
monotremes and marsupials. J. Mol. Evol., v. 45, pp. 311–321, 1997.
Lanave, C.; Colangelo, A..; Saccone, C.; Alberghina, L. Molecular evolution of the neurotrophin family
members and their Trk receptors. Gene, v.394, pp. 1-12, 2007.
Lanktree, M.; Squassina, A.; Krinsky, M. et al. Association Study of Brain-Derived Neurotrophic Factor
(BDNF) and LIN-7 Homolog (LIN-7) Genes with Adult Attention-Deficit/Hyperactivity Disorder. Am J
Med Genet B Neuropsychiatr Genet., v.147, n. 6, pp. 945-51, 2008.
Lebmann, V.; Brigadski, T. Mechanisms, locations, and kinetics of synaptic BDNF secretion: An
update. Neurosci Res, v. 65, pp. 11–22, 2009.
Lee, R.; Kermani, P.; Teng, K.; Hempstead, B. Regulation of cell survival by secreted
proneurotrophins. Science, v. 294, pp.1945–48, 2001.
Leibrock, J.; Lottspeich, F.; Hohn, A. et al. Molecular cloning and expression of brain derived
neurotrophic factor. Nature, v.341, n.6238, pp.149-52, 1989.
Lou, H.; Kim, S-K; Zaitsev, E. et al. Sorting and Activity-Dependent Secretion of BDNF Require
Interaction of a Specific Motif with the Sorting Receptor Carboxypeptidase E. Neuron, v. 45, n. 2, pp.
245-55, 2005.
Lüthy, R.; Bowie, J.; Eisenberg, D. Assessment of protein models with three-dimensional profiles.
Nature, v. 356, pp. 83 -85, 1992.
Martí-Renom, M.; Stuart, A.; Fiser, A. et al. Comparative Protein Structure Modeling of Genes and
Genomes. Annu. Rev. Biophys. Biomol. Struct, v. 29, pp. 291 – 325. 2000.
Matsumoto, T.; Rauskolb, S.; Polack, M. et al. Biosynthesis and processing of endogenous BDNF:
CNS neurons store and secrete BDNF, not pro-BDNF. Nat Neurosci., v. 11, n. 2, pp. 131-33, 2008.
Mercader, J.; Ribasés, M.; Gratacòs, M. et al. Altered brain-derived neurotrophic factor blood levels
and gene variability are associated with anorexia and bulimia. Genes, brain and behavior, v.6, pp.
706-16, 2007.
Miller-Butterworth, C.; Murphy, W.; O'Brien, S. et al. A Family Matter: Conclusive Resolution of the
Taxonomic Position of the Long-fingered Bats, Mniopterus, Mol. Biol. Evol., v. 24, n.7, pp.1553-61,
2007.
Mowla, S.; Farhadi, H.; Pareek, S. et al. Biosynthesis and post-translational processing of the
precursor to Brain-Derived Neurotrophic Factor. J. Biol. Chem., v. 276, n. 16, pp. 12660-66, 2001.
Murphy, W.; Eizirik, E.; O´Brien, S. et al. Resolution of the early placental mammal radiation using
Bayesian phylogenetics. Science, v. 294, n. 5550, pp. 2348-51, 2001.
Nagappan, G.; Zaitsev, E.; Senatorov, V. et al. Control of extracellular cleavage of proBDNF by high
frequency neuronal activity. Proc. Natl. Acad. Sci. USA, v. 106, n.4, pp. 1267-72, 2009.
Nayeem, A.; Sitkoff, D.; Junior, S. A comparative study of available software for high accuracy
homology modeling: from sequence alignments to structural models. Protein Sci., v.15, pp.808-24,
2006.
Nomoto, H.; Takaiwa, M.; Mouri, A.; Furukawa, S. Pro-region of neurotrophins determines the
processing efficiency. Biochem Biophys Res Commun, v. 356, n.4, pp. 919 – 24, 2007.
Paiardini, A.; Caputa, V. Insights into the interaction of sortilin with proneurotrophins: A computational
approach. Neuropeptides, v. 42, pp. 205 – 14, 2008.
69
Pang, P.; Teng, H.; Zaitsev, E. et al. Cleavage of proBDNF by tPA/Plasmin is essential for long-term
hippocampal plasticity. Science, v. 306, n. 5695, pp. 487- 91, 2004.
Paoletti, F.; Covaceuszach, S.; Konarev, P. et al. Intrinsic structural disorder of mouse proNGF.
Proteins, v. 75, pp. 990 – 1009, 2009.
Park, A.; Holmer, L.; Worman, H. A human HP1 pseudogene maps to chromosome 11p14. Somat Cell
Mol Genet., v.24, n.6, pp. 353-6, 1998.
Pearlman, D.; Case, D; Caldwell, J. et al. AMBER, a package of computer programs for applying
molecular mechanics, normal mode analysis, molecular dynamics and free energy calculations to
simulate the structural and energetic properties of molecules. Comp. Phys. Commun, v. 91, pp. 1-
41, 1995.
Pruunsild, P.; Kazantseva, A.; Aid, T. et al. Dissecting the human BDNF locus: bidirectional
transcription, complex splicing, and multiple promoters. Genomics, v. 90, n.3, pp.397-406, 2007.
Rattenholl, A.; Lilie, H.; Grossmann, A. et al. The pro-sequence facilitates folding of human Nerve
Growth Factor from Escherichia coli inclusion bodies. Eur. J. Biochem., v. 268, pp. 3296 -3303,
2001b.
Rattenholl, A.; Ruoppolo, M.; Flagiello, A. et al. Pro-sequence assisted folding and disulfide bond
formation of human Nerve Growth Factor. J. Mol. Biol., v. 305, pp. 523 – 33, 2001a.
Reichardt, L. Neurotrophin-regulated signalling pathways. Phil. Trans. R. Soc. B, v.361, pp. 1545–64,
2006.
Robinson, R..; Radziejewski, C.; Stuart, D.; Jones, E. Structure of the Brain-Derived Neurotrophic
factor/ Neurotrophin 3 Heterodimer. Biochem., v. 34, pp. 4139-45, 1995.
Robinson, R.; Radziejewski, C.; Spraggon, G; et al. The structures of the neurotrophin 4 homodimer
and the brain-derived neurotrophic factor/neurotrophin 4 heterodimer reveal a common Trk-binding
site. Protein Sci., v.8, pp. 2589-97, 1999.
Sali, A.; Blundell, T. Comparative Protein Modelling by Satisfaction of Spatial Restraints. J. Mol. Biol,
v.234, n.3, pp. 779 – 815, 1993.
Schwede, T.; Kopp, J.; Guex, N.; Peitsch, M. Swiss-Model: an automated protein homology-moleling
server. Nucl Ac Res, v. 31, n. 13, pp. 3381-85, 2003.
Seidah, N.; Mowla, S.; Hamelin, J. et al. Mammalian subtilisin / Kexin isozyme SKI-1: A widely
expressed proprotein convertase with a unique cleavage specificity and cellular localization. Proc.
Natl. Acad. Sci. USA, v. 96, pp. 1321 – 26, 1999.
Teng, H.; Teng, K.; Lee, R. et al. ProBDNF induces neuronal apoptosis via activation of a receptor
complex of p75NTR and sortilin, J. Neurosci., v. 25, pp.5455–63, 2005.
Van Den Bussche, R.; Hoofer, S. Phylogenetic relationships among recent chiropteran families and
the importance of choosing appropriate out-group taxa. J Mammal, v. 85, pp. 321–30, 2004.
Wallace, A.; Laskowski, R.; Thornton, J. LIGPLOT: a program to generate schematic diagrams of
protein-ligand interactions. Protein Eng. Des, Sel. 8, pp. 127 –34, 1995.
Woo, N.; Teng, H.; Siao, T-J., et al. Activation of p75NTR by proBDNF facilities hippocampal long-
term depression. Nat neurosc., v. 8, pp. 1069-77, 2005.
70
Yamada, K.; Mizuno, M.; Nebeshima, T. Role for brain-derived neurotrophic factor in learning and
memory. Life Sci., v.70, n.7, pp. 735-44, 2002.
Yang, J.; Sião, C-J; Nagappan, G., et al. Neuronal release of pro-BDNF. Nat Neurosci, v.12, n.2, pp.
113-15, 2009.
71