Article

Evolution of High Mobility Group Nucleosome-Binding Proteinsand Its Implications for Vertebrate Chromatin SpecializationRodrigo Gonzalez-Romeroy1 Jose M Eirın-Lopezy2 and Juan Ausio1

1Department of Biochemistry and Microbiology University of Victoria Victoria BC Canada2Chromatin Structure and Evolution (CHROMEVOL) Group Department of Biological Sciences Florida International UniversityyThese two authors have contributed equally to this work

Corresponding author E-mail jausiouvicca

Associate editor Helen Piontkivska

Abstract

High mobility group (HMG)-N proteins are a family of small nonhistone proteins that bind to nucleosomes (N) Despitethe amount of information available on their structure and function there is an almost complete lack of information onthe molecular evolutionary mechanisms leading to their exclusive differentiation In the present work we provideevidence suggesting that HMGN lineages constitute independent monophyletic groups derived from a common ancestorprior to the diversification of vertebrates Based on observations of the functional diversification across vertebrate HMGNproteins and on the extensive silent nucleotide divergence our results suggest that the long-term evolution of HMGNsoccurs under strong purifying selection resulting from the lineage-specific functional constraints of their different proteindomains Selection analyses on independent lineages suggest that their functional specialization was mediated by burstsof adaptive selection at specific evolutionary times in a small subset of codons with functional relevancemdashmost notablyin HMGN1 and in the rapidly evolving HMGN5 This work provides useful information to our understanding of thespecialization imparted on chromatin metabolism by HMGNs especially on the evolutionary mechanisms underlyingtheir functional differentiation in vertebrates

Key words chromatin high mobility group proteins HMGN nucleosome-binding domain long-term evolution purifyingselection episodic adaptive selection

IntroductionThe high mobility group (HMG) proteins are the most abun-dant and ubiquitous nonhistone chromosomal proteinsThey bind to DNA and to nucleosomes eliciting structuralchanges on DNA metabolic activities such as transcriptionreplication and DNA repair (Bustin and Reeves 1996 Bustin1999) The HMG superfamily is composed of three familiesHMGA HMGB and HMGN (Bustin 2001a) each containing aunique structural motif (Bianchi and Agresti 2005) The char-acteristic domains are AT-hook for the HMGA family theHMG Box for the HMGB family and the nucleosome-bindingdomain (NBD) for the members of the HMGN family

The first HMGN proteins (HMGN1 and HMGN2) wereidentified by EW Johns group under the names HMG-14and HMG-17 (Johns 1982) (see [Bustin 2001b] for a changein HMG nomenclature) HMGNs are expressed only in verte-brates and interact with the 145-bp nucleosome core particle(NCP) (fig 1A) without any DNA sequence specificity (Bustin2001a) Binding of HMGNs to nucleosomes has downstreamfunctional implications for transcription replication andrepair (Vestner et al 1998 Bustin 2001a Birger et al 2005Belova et al 2008 Kim et al 2009 Postnikov and Bustin 2010Zhu and Hansen 2010) Nucleosome binding is mediated bythe NBD a highly conserved 30 amino acid domain in verte-brates (Bustin 2001a) (fig 1A) It contains an eight amino acidmotif RRSARLSA responsible for the anchoring of HMGNs tothe NCP (Ueda et al 2008) Sumoylation of lysines 17 and 35

in HMGN2 within the NBD region has been recently shownto decrease the nucleosome-binding affinity of these proteins(Wu et al 2014) In addition to the NBD all HMGN proteinshave a bipartite nuclear localization signal (Hock Scheer et al1998) and a negatively charged regulatory domain (RD) intheir C-terminal region that mediates chromatin unfolding(Bustin 2001a) and plays an important role in the effects ofHMGNs on histone posttranslational modifications (PTMs)(Ueda et al 2006) The HMGN family consists of five closelyrelated proteins that have been detected only in vertebratesHMGN1 HMGN2 HMGN3 HMGN4 and HMGN5Although HMGN1 to HMGN4 are approximately 100amino acids in length HMGN5 is much larger (300ndash400amino acids) due to the presence of a long acidic C-terminalregion (Rochman et al 2010) that affects the cellular locali-zation and architectural properties of the protein (Rochmanet al 2009)

HMGN1 and HMGN2 are the most abundant and char-acterized proteins of the HMGN family In addition to nucle-osome binding they reduce the compactness of chromatinfiber and enhance transcription of chromatin templates(Bustin 2001a) Both in vivo and in vitro studies suggestthat they bind to NCPs forming homodimeric complexescontaining two molecules of either HMGN1 or HMGN2(Postnikov et al 1995 1997) Using cross-linking andmethyl-based NMR analysis it has been demonstrated thatthese proteins unfold chromatin by targeting several of the

The Author 2014 Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution All rights reserved For permissions pleasee-mail journalspermissionsoupcom

Mol Biol Evol 32(1)121ndash131 doi101093molbevmsu280 Advance Access publication October 3 2014 121

at Florida International University on D

ecember 22 2014

httpmbeoxfordjournalsorg

Dow

nloaded from

main elements essential to chromatin compaction the linkerhistone H1 and the N-terminal tails of histones H3 and H4(Trieschmann et al 1998 Catez et al 2002 Kato et al 2011) Inaddition it has also been shown that both HMGN1 andHMGN2 can form multiple complexes with other nuclearproteins which could alter their chromatin interaction andbiological function (Lim et al 2002 Kugler et al 2012)

Although HMGN1 and HMGN2 seem to be highly ex-pressed during embryogenesis they are also ubiquitously ex-pressed in several adult tissues (Furusawa and Cherukuri2010) Immunofluorescence studies have pinpointed theirnumerous foci within the nucleus (Postnikov et al 1997)Such organization appears to be highly dynamic and depen-dent on transcriptional activity (Hock Wilde et al 1998)Binding of HMGN1 and HMGN2 proteins to nucleosomesis affected by PTMs such as phosphorylation and acetylationthat can reduce or even abolish their nucleosome-bindingability (Bergel et al 2000 Prymakowska-Bosak et al 2001Gerlitz et al 2009 Pogna et al 2010) Finally several studiesalso suggest that both proteins can indirectly modulate thelevels of some of the histone PTMs hence affecting the his-tone-mediated epigenetic regulation of gene expression(Postnikov and Bustin 2010 Kugler et al 2012)

HMGN3 (formerly Trip7 [thyroid hormone receptor inter-acting protein 7]) is the only family member to consist of twosplice variants HMGN3a and HMGN3b (West et al 2001)The shorter HMGN3b variant lacks the C-terminal RD Duringthe course of this study while searching in NCBI databasestwo more splice variants were identified we have namedthem HMGN3c and HMGN3d (see Results and Discussionsection and supplementary table S1 Supplementary Materialonline) It remains to be determined whether any of theseHMGN3 variants play a distinct role in vivo AlthoughHMGN1 and HMGN2 are ubiquitously expressed and

involved in general cellular differentiation HMGN3 expres-sion seems to be tissue-specific and dependent upon devel-opment (Ueda et al 2009) In mouse and human tissuesHMGN3 is highly expressed in the eye and in the brain(West et al 2001 2004 Ito and Bustin 2002) where itmight play a role in astrocyte function (Ito and Bustin2002) Furthermore HMGN3 is also abundant in adult pan-creatic islet cells where it modulates the transcriptional pro-gram of these cells affecting insulin secretion (Ueda et al2009)

HMGN4 is the least-studied member of all HMGNsClosely related to HMGN2 HMGN4 was identified in 2001during a GenBank database-search of a new HMGN2-liketranscript (Birger et al 2001) In contrast to the rest ofHMGNs which are all encoded by genes containing six dis-tinct exons HMGN4 is encoded by an intronless gene (Birgeret al 2001) Also although all the other HMGNs have beendetected in all vertebrates that have been tested the genecoding for HMGN4 appears to be restricted to primates(Kugler et al 2012) The HMGN4 gene seems to have origi-nated around 25 Ma from a fortuitous insertion of anHMGN2 retro-pseudogene next to an active promoter(Birger et al 2001) This had been earlier recognized as apossibility (Srikantha et al 1987) as HMGN is one of thelargest known retro-pseudogene families with human andmouse genomes containing more than 30 retro-pseudogenesfor HMGN1 and HMGN2 dispersed over many chromo-somes (Popescu et al 1990 Johnson et al 1992 1993Strichman-Almashanu et al 2003) HMGN4 expression ap-pears to be widespread among human tissues (with ahigher expression in the thyroid gland thymus andlymph nodes) albeit with a transcript and protein abun-dance significantly lower than that of HMGN2 (Birger et al2001)

FIG 1 Schematic representation of the interactions of HMGNs with chromatin (A) Interaction of HMGN2 with the nucleosome (Kato et al 2011Kugler et al 2012) The core histones are depicted in different light colors H3 blue H4 green H2A yellow H2B pink The green oval indicates theapproximate location of the acidic patch (Luger et al 1997) The colors for the HMGN2 molecule correspond to the different structural regions along itsamino acid sequence (as indicated in fig 2) Interaction of the NBD of HMGNs with the nucleosome positions their C-terminal domain near thenucleosome dyad This results in an impairment of the proper binding of the winged histone domain (WHD) (Kasinsky et al 2001) of linker histones tothis region (Zhou et al 2013) (B) Interaction of HMGN5 with chromatin results in a relaxed open conformation of the chromatin fiber which preventshistone H1 binding Such an unfolding stems from the binding competition between HMGN5 and histone H1 for the dyad region of the nucleosomeandor from the juxtaposition of their respective negatively and positively charged C-terminal domains (Rochman et al 2009 2010) The red circle in thehistone H1 molecule represents its highly characteristic WHD The double arrow underscores the highly dynamic nature of the interactions of histoneH1 and HMGNs with the chromatin template (Kugler et al 2012)

122

Gonzalez-Romero et al doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

HMGN5 (previously known as NBP-45 [nucleosomal-bind-ing protein 45] or NSBP1 [nucleosome-binding protein 1]) isthe most recently described HMGN variant (Shirakawa et al2000 King and Francomano 2001 Rochman et al 2009) It is arapidly evolving protein that modulates the dynamic bindingof linker histones (histone H1) to chromatin reducing thecompaction of the chromatin fiber and affecting transcrip-tion HMGN5 contains a long acidic C-terminal domain thatdiffers among different vertebrate species (Malicet et al 2011)The exon that encodes for this C-terminal region (exon VI)contains sequences highly similar to both HAL1 retro-trans-posable element and HERVH endogenous retrovirus thesesimilarities could be related to HMGN5rsquos rapidly evolvingnature (King and Francomano 2001 Malicet et al 2011)The C-terminal region of HMGN5 is the main determinantof its chromatin interaction properties and its chromatin lo-cation (Rochman et al 2009) For instance mouse HMGN5(with a 300 amino acid C terminus) (fig 1B) is preferentiallyfound in euchromatin whereas human HMGN5 (with a 200amino acid C terminus) exhibits a less restricted dual euchro-matin and heterochromatin localizationmdashsimilar to otherHMGN variants (Malicet et al 2011) Although its biologicalfunction is unknown overexpression of HMGN5 has beenobserved in several human tumors such as in prostatecancer (Jiang et al 2010) squamous cell carcinoma (Greenet al 2006) renal cell carcinoma (Ji et al 2012) breast cancer(Li et al 2006) gliomas (Qu et al 2011) and lung cancer (Chenet al 2012) this suggests that HMGN5 plays a role in tumor-igenesis Knockdown of HMGN5 induces cell cycle arrest andapoptosis in these human tumor cell lines it has thus beensuggested that HMGN5 might be a potential molecular targetfor cancer therapy (Chen et al 2012)

In the present work we trace the phylogeny and evolu-tionary history of HMGNs which led to their structural andfunctional specialization during the course of vertebrateevolution

Results and Discussion

Vertebrate HMGN2 Distribution and Tissue Variation

As mentioned in the introduction HMGN12 display a het-erogeneous pattern of distribution and expression across ver-tebrates an animal group within which they appear to havehad their evolutionary emergence Attempts to extract anysimilar proteins in invertebrate organisms were unsuccessfulin both this and in previous studies (Bustin 2001a) Figure 2Ashows the alignment of HMGN1 and HMGN2 in five speciesrepresentative of each of the five classes within the subphy-lum vertebrata The Logos representation shown underneaththe amino acid sequences highlights the extent to which theirdifferent structural domains have been conserved

To gain insight regarding the distribution of these HMGNsand their relative abundance a 5 perchloric acid (PCA) ex-traction was performed on a liver sample from the sameorganisms used in the sequence alignments This type ofacid extraction not only extracts HMGNs but it also extractsthe linker histones of the histone H1 family (Goodwin et al1978) We took advantage of this dual extraction to produce

an approximate normalization of the protein loadings foreach extraction (fig 2B) prior to performing the Western-blot analysis with an HMGN2 mouse antibody Attempts toperform a similar analysis with HMGN1 and HMGN3 mouseantibodies proved to be completely unsuccessful As shown infigure 2B HMGN2 exhibits a variable distribution across thevertebrate species analyzed here with a lower expression inchicken and an enhanced electrophoretic mobility inzebrafishmdashin agreement with the smaller size of the aminoacid sequence in this organism (see fig 2A and supplementaryfig S1 Supplementary Material online)

More striking is the variability that is observed across dif-ferent tissues within the same organism as exemplified by theWestern-blot analysis carried out on mice and shown infigure 2C The major occurrence of HMGN2 appears to bein the brain followed by intestines and lungs testes kidneysand liver In partial agreement with these results a previousstudy on the variation of HMGN2 in liver kidney and lungtissues of rats was not able to detect a significant variationwithin these tissues in this organism but did consistentlyshow a larger presence in lung tissue (Kuehl et al 1984)Although the presence of HMGN2 in the nucleus has beenrelated to the transcriptional activity of the cell (Hock Wildeet al 1998) its relation to the tissue variability observed byusmdashand its potential significancemdashdeserves furtherattention

As mentioned above attempts to extend our tissue andorganism distribution analysis using mouse antibodies failedFigure 3 provides an amino acid sequence analysis for thisHMGN3 as well as for the primate-specific HMGN4 andHMGN5 In the absence of a Western-blot analysis tool ourbioinformatics search for HMGN3 occurrence only allowed usto detect the presence of this protein in mammals in birdsand in Xenopus but not in any other vertebrate species forwhich whole-genome information is available

Phylogenetic Relationships among HMGN FamilyMembers

The availability of complete information on many vertebrateorganism genomes provides a unique opportunity to addressa fundamental question as to how the different HMGNscorrelate to each other throughout vertebrate evolution Tothis end protein and gene phylogenies were reconstructedfrom sequence data obtained after exhaustive moleculardata mining (see supplementary table S1 SupplementaryMaterial online) The resulting protein and gene phylogeniesare shown in figure 4 and supplementary figure S2Supplementary Material online respectively In both in-stances the five major HMGN lineages as well as theHMG-14A group are well defined each of them representsa distinctive monophyletic clade as supported by the highconfidence values observed Given that the bootstrap (BS)method is known to be conservative values higher than80 were interpreted as high statistical support for internalnodes on both trees The results were additionally supportedby high Bayesian posterior probabilities in those branchesleading to each HMGN lineage Such a clustering pattern is

123

Long-Term Evolution of HMGN Proteins doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

consistent with the presence of specific constraints actingupon different HMGN lineages which leads to a functionaldiversification that is likely to have different downstreamstructural and functional implications for chromatin

The reconstructed topologies support a retroviral origin ofHMGN4 (from an HMGN2 retro-pseudogene [Birger et al2001]) as well as a close relationship between the birdreptileHMG-14A group and HMGN3 (Browne and Dodgson 1993)(table 1) Unfortunately actual sequence data does not allowus to indicate which HMGN is the closest one to a commonancestor Although the confidence values obtained for inter-nal nodes within the protein phylogeny allow us to discernbeween monophyletic groups corresponding to each HMGN

type it is not possible to solve the deep relationships amongHMGNs beyond each group probably due to the accumula-tion of multiple substitutions at individual amino acid sitesHowever the taxonomic distribution and the wide distribu-tion of HMGNs across vertebrates suggest that HMGN1 andHMGN2 (the two founding members of the HMGN family) aswell as HMGN3 arose earlier in evolution In contrastHMGN4 (present in catarrhini primates) and HMGN5 (pre-sent in mammals) appear to be the most recent lineagesoriginating 25 and 300 Ma respectively (Birger et al 2001Malicet et al 2011) with the latter corresponding to themost sequence-divergent as a result of its rapid evolution(Malicet et al 2011) (table 1)

FIG 2 HMGN1 and HMGN2 (A) Protein sequence alignment for a representative organism of each of the five vertebrate classes Zebrafish Danio rerio(fish) African clawed frog Xenopus laevis (amphibian) Carolina anole Anolis carolinensis (reptile) chicken Gallus gallus (bird) and mouse Musmusculus (mammal) The combined Logos representations using alignments from supplementary figure S1 Supplementary Material online are alsoshown (B) Western-blot analysis of HMGN2 from liver tissue-PCA extracts from each one of the vertebrate representatives in (A) A coomassie blue-stained replica SDSndashPAGE corresponding to the histone H1 fraction coextracted in this way is also shown (C) Coomassie blue stained SDSndashPAGE andWestern-blot analysis of HMGN2 PCA extracted from different mouse tissues (liver brain testis kidney lung and gut) In (B) and in (C) histones H1were used for protein loading normalization purposes

124

Gonzalez-Romero et al doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

Mechanisms of HMGN Evolution

The phylogenetic analysis shown in figure 4 depicts a highlyspecialized differentiation of HMGNs which is likely related toa functional specialization Which then are the mechanismsthat govern the long-term evolution of these different line-ages To address this question we started by examining theprotein variation within each of the different HMGN lineagesSuch analysis revealed that HMGN5 is the most diverse(p = 0326 0015) followed by HMGN1 HMGN3 andHMGN2 with HMGN4 having the lowest levels of variation(p = 0004 0004) (table 2) The nature of the nucleotidevariation underlying such diversity was predominantly synon-ymous and in all instances higher than the nonsynonymousvariation As expected the lowest levels of silent variationwere found in HMGN4 (pS = 0055 0020) likely mirroringits recently retroposed origin (Birger et al 2001 Strichman-Almashanu et al 2003) Still when it comes to completeproteins codon-based Z-tests for selection consistently re-vealed significant differences between synonymous andnonsynonymous variation (table 2) Altogether these resultssupport the presence of a strong purifying selection operatingon the different HMGN protein lineagesmdashwhich is most likelyresponsible for preserving the structural features required forthe specific interaction of each of these proteins with thenucleosome (Bustin 2001a)

Evidence for the role of purifying selection was furthersupported by the low levels of protein variation found atthe N-terminal domain of HMGNs (table 2) This region prob-ably represents the main target of selection as it encompassesthe most functionally relevant binding domain (NBD) for theinteraction of these proteins with the nucleosome (Kato et al2011) Comparatively the higher nonsilent variation found atthe C-terminal region is probably due to a low selectivity foracidic amino acids in the regulatory domain (RD) This be-comes especially evident in the long C-terminal region ofHMGN5 which contains high levels of either aspartic or glu-tamic acid organized in the repetitive motif EDGKE Thehighly acidic nature of this domain represents the main de-terminant of the chromatin interaction properties of HMGN5(fig 1B) (Malicet et al 2011) and it also plays a critical role intranscriptional regulation by modulating the occurrence ofspecific chromatin modifications (Ueda et al 2006)

To test whether HMGN specialization hints at the involve-ment of additional lineage-specific functional constrains weestimated the pace at which each HMGN lineage evolves Theanalysis showed low-to-moderate rates of evolution in all in-stancesmdashexcept in HMGN5 which appears to be evolving ata very fast rate (fig 5) In this regard HMGN5 represents alineage with an outstandingly rapid rate of evolution remi-niscent of chromosomal reproductive proteins (Eirın-Lopez

FIG 3 HMGN3 HMGN4 and HMGN5 Protein sequence alignment for different representative organisms chicken Gallus gallus cow Bos taurusmouse Mus musculus Rhesus macaque Macaca mulatta chimpanzee Pan troglodytes orangutan Pongo abelii human Homo sapiens The combinedLogos representations using alignments from supplementary figure S1 Supplementary Material online are also shown

125

Long-Term Evolution of HMGN Proteins doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

et al 2008 Ishibashi et al 2010) Quite unexpectedly such ahigh rate of evolution does not preclude the use of preferredcodons in HMGN5 genes as shown by the codon bias

estimations (table 2) This would support the existence ofspecialized constraints in the evolution of HMGN5 whichare different from those operating in other lineages

Episodic Selection within HMGN Lineages

Despite all the consistent evidence for the importance ofpurifying selection in shaping the functional differentiationof HMGNs the presence of heterogeneous evolutionaryrates across lineagesmdashtogether with the high level of diver-gence displayed by the recently differentiated HMGN5 line-agemdashraises the question as to whether or not there is anyevidence for adaptive selective episodes driving the rapid dif-ferentiation of specific HMGN lineages Should this be thecase it would be expected to have traces of these episodesdetected across HMGN evolution This notion is supportedby our results which show a significant departure from aglobal clock-like behavior during the evolution of HMGNproteins (lnL without clock =59980 lnL with clock-=213044 Plt 0001) resulting from heterogeneous ratesof evolution at internal branches leading to the differentHMGN lineages (Plt 0001 fig 4)

Because HMGN5 lineage is only present in mammals wedecided to base our analysis on the evolution of HMGN genesin this group Lineages HMGN2 HMGN4 and HMGN3 areclosely related within a single monophyletic groupmdashwithHMGN1 and the HMGN5 lineages constituting a sisterclade (fig 6A) As it was done for vertebrates the global mo-lecular clock hypothesis was also tested and rejected in themammalian groups (Plt 0001) which exhibit a significantdeparture from a clock-like behavior found at the monophy-letic origins of each HMGN clade (fig 6A) Given the presenceof heterogeneous rates of evolution we investigated to whatextent those resulted from specific selective episodes operat-ing on particular HMGNs The screening of the HMGN phy-logeny revealed significant traces of episodic adaptiveselection (4 1) on at least five internal branches(P 005) (fig 6A) Although one of these branches is locatedat the root of the HMGN4 lineage the four remainingbranches are located in the subtree encompassing lineagesHMGN1 and HMGN5 including the root of this clade(P 001) as well as the internal branch leading to theHMGN1 lineage (P 001) and the branches groupingmurine (P 0001) and catarrhini (P 005) HMGN5 genestogether

FIG 4 Phylogenetic maximum likelihood (ML) relationships amongvertebrate HMGN protein lineages The numbers for interior branchesrepresent nonparametric bootstrap (BS) probabilities based on 1000replications followed by Bayesian posterior probabilities (only shownwhen BS 50 or posterior probability 05) Two black circles at in-ternal nodes indicate subtrees at which the molecular clock hypothesiswas rejected (Plt 0001) after testing for the presence of local molecularclocks

Table 1 Evolutionary Divergence between HMGN Protein Lineagesacross Vertebratesa

HMGN1 HMGN2 HMGN3 HMGN4 HMGN5

HMGN1 mdash

HMGN2 249 58 mdash

HMGN3 282 63 220 63 mdash

HMGN4 310 65 88 44 247 69 mdash

HMGN5 337 72 381 73 325 68 383 74 mdash

aAverage amino acid substitutions per 100 sites (p-distance) Standard errors werecalculated using the BS method with 1000 replicates

126

Gonzalez-Romero et al doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

Additional insight was gained by combining maximumlikelihood (ML) and Bayesian selection analyses which al-lowed us to disclose the individual sites subject to diversifyingselection (Kosakovsky Pond and Frost 2005) As a result 12positively selected and 134 negatively selected sites were iden-tified based on the consensus of single-likelihood ancestorcounting (SLAC) fixed effects likelihood (FEL) random effectslikelihood (REL) and fast unconstrained Bayesian approxima-tion (FUBAR) methods (table 3) Among them seven posi-tively selected codons were consistently identified as subjectto episodic positive selection based on the mixed effectsmodel of evolution (MEME) method (Plt 01) includingthree positions common to all HMGN types (49 53 and97) and four positions exclusively from the long C-terminalregion of the HMGN5 lineage (table 3 and fig 6B) The phy-logenetic analysis of the mutations at these positions suggeststhat changes at codons 53 and 97 were most likely involved in

the differentiation of the HMGN1 lineage with changes inpositions 49 and 53 linked to HMGN5 Interestingly the pres-ence of episodic selection at position 49 could constitute amajor driver of HMGN5 specialization given the location ofthis codon within the highly conserved and functionally rel-evant NBD region Nonetheless the differentiation of thislatter lineage also required additional substitutions at posi-tions 135 363 431 and 433 (fig 6B)

ConclusionsHMGNs are characterized by their heterogeneous pattern ofdistribution and expression across vertebrates and have crit-ical functions in chromatin metabolism Yet the evolutionarymechanisms responsible for such diversification and for thefunctional differentiation across their family members haveeluded study In the present work we provide the first com-prehensive analysis of the evolution of HMGNs supplyingevidence for three previously unknown major findings 1)phylogenetic relationships among HMGN lineages showthat all of them are independent monophyletic groups arisingfrom a common ancestor that preceded the diversification ofvertebrates 2) long-term evolution of HMGNs is predomi-nantly driven by purifying selection resulting from lineage-specific functional constraints of their different proteindomains 3) functional specialization of the different HMGNlineages occurred by bursts of adaptive selection at specificevolutionary times and protein positions most notably inHMGN1 and in the rapidly evolving HMGN5 Altogetherour results suggest that HMGN evolution involves a hetero-geneous process largely shaped by strong purifying selectionwith occasional episodes of diversifying selection geared to-ward the functional specialization of the different lineages

Table 2 Average Numbers of Amino Acid (pAA) Nucleotide (pNT) Synonymous (pS) and Nonsynonymous (pN) Nucleotide Differences per 100Sites Site in HMGN Lineages Discriminating among Complete Coding Regions N-terminal and C-Terminal Domainsa

HMGN Type pAA (SE) pNT (SE) pS (SE) pN (SE) R Z-test ENC

HMGN1 complete 233 24 221 14 499 24 130 15 10 137 501 59

HMGN1 N-terminus 164 36 209 21 530 25 99 21 08 134 517 107

HMGN1 C-terminus 285 34 229 18 471 34 153 20 12 80 446 87

HMGN2 complete 68 13 107 10 304 25 35 07 17 101 453 76

HMGN2 N-terminus 63 18 114 14 325 34 34 09 16 82 488 62

HMGN2 C-terminus 75 19 98 15 273 38 36 10 17 58 497 00

HMGN3 complete 87 18 96 10 229 23 43 09 15 77 431 57

HMGN3 N-terminus 77 20 95 11 245 26 38 10 14 75 496 66

HMGN3 C-terminus 104 30 97 17 200 41 51 15 17 31 431 78

HMGN4 complete 04 04 15 05 55 20 02 02 07 26 470 13

HMGN4 N-terminus 00 00 17 08 66 30 00 00 10 22 425 00

HMGN4 C-terminus 29 09 12 07 39 26 04 04 06 12 398 00

HMGN5 complete 326 15 194 08 231 016 183 10 14 27 399 23

HMGN5 N-terminus 162 35 98 17 139 039 83 19 17 12 349 64

HMGN5 C-terminus 356 17 212 08 252 017 200 11 14 27 386 34

NotemdashSE standard error ENC Effective Number of Codons (codon bias) ranging between 61 (no bias) and 20 (maximum bias) HMGN1 N-terminus nucleotide positions 1ndash153 C-terminus positions 154ndash342 HMGN2 N-terminus positions 1ndash147 C-terminus positions 148ndash279 HMGN3 N-terminus positions 1ndash147 C-terminus positions 148ndash396HMGN4 N-terminus positions 1ndash141 C-terminus positions 142ndash273 HMGN5 N-terminus positions 1ndash126 C-terminus positions 127ndash1314 (see Materials and Methods section fora detailed explanation)aThe average transitiontransversion ratio used in the estimation of pS and pN is denoted as R SEs calculated by the bootstrap method with 1000 replicates

Plt 005 and Plt 0001 level in Z-test comparisons (pS 4 pN)

FIG 5 Estimated rates of evolution for HMGN proteins Evolutionaryrates for the fast-evolving chromosomal proteins histone H2ABbd aswell as histone H1 and histones H2AH2B (dashed lines) are included asreference HMGN4 is not shown due to its very slow rate of evolution

127

Long-Term Evolution of HMGN Proteins doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

Materials and Methods

Extraction and Analysis of Distribution of HMGNProteins

HMGN proteins were isolated from liver tissue of differentvertebrate representatives including Fish (zebrafish Daniorerio) amphibian (African clawed frog Xenopus laevis) reptile(Carolina anole Anolis carolinensis) bird (chicken Gallusgallus) and mammalian (mouse Mus musculus) representa-tives In addition HMGNs were also extracted from severalmouse tissues including brain testis kidney lung and intes-tinemdashas described elsewhere (Lim et al 2004) Briefly thetissues were processed with a dounce homogenizer in015 M NaCl 10 mM Tris-HCl (pH 75) and a 05 TritonX-100 buffer containing Roche Complete Protease cocktailinhibitor (Roche Molecular Biochemicals Laval QC) at aratio of 1100 vv After homogenization and incubation onice for 5 min the samples were centrifuged at 12000 g for10 min at 4 C The resulting pellets were resuspended in 5PCA homogenized as above and centrifuged in the sameway 1 N HCl was added to the PCA supernatant extractsto bring the solution to 02 N HCl Then the PCA supernatantextracts were precipitated with six volumes of acetoneat 20C overnight and centrifuged at 12000 g for10 min at 4 C The acetone pellets were dried using a speed-vac concentrator and stored at 80C until further use inpolyacrylamide gel electrophoresis (PAGE) and Western-blotanalyses

Gel Electrophoresis and Western Blotting

Sodium dodecyl sulfate (SDS)ndashPAGE (15 acrylamide 04bis-acrylamide) was carried out using the approach describedby Laemmli (Laemmli and Johnson 1973) Western-blot anal-yses were performed using a mouse anti-HMGN2 antibody (agenerous gift from Michael Bustin) Gels were electro-trans-ferred to a polyvinylidene difluoride membrane (Bio-RadHercules CA) and processed as described elsewhere (Finnet al 2008) HMGN2 antibody was used at a 12000 dilutionMembranes were incubated with secondary goat antirabbitantibody (GE Healthcare Baie drsquoUrfe QC) at a 15000 dilu-tion Secondary antibody was detected with enhanced chemi-luminescence (GE Healthcare) and exposure to X-ray films

Molecular Data Mining

Extensive data mining experiments were performed inthe GenBank database (wwwncbinlmnihgovgenbank) inorder to collect all the HMGN sequences available as ofJanuary 2014 Altogether 88 nt coding sequences belongingto 21 different vertebrate species were used in the presentwork including 18 HMGN1 20 HMGN2 33 HMGN3 5HMGN4 9 HMGN5 3 HMG-14A and 1 outgroup sequence(HMGA1 from human see supplementary table S1Supplementary Material online) Sequences were revised forerrors in accession numbers and nomenclature and giventhat the HMGN family is one of the largest known retro--pseudogene families (Strichman-Almashanu et al 2003) onlyfunctional HMGN coding sequences were selected Multiple

FIG 6 Selection episodes involved in the evolution of mammalian HMGN lineages (A) ML gene tree depicting episodes of diversifying selection duringHMGN differentiation in mammals Numbers for interior branches are indicated as in figure 4 Deviations from the molecular clock at internal subtreesare indicated by one (Plt 001) or two (Plt 0001) black circles at the corresponding internal braches The strength of selection at significant branches isrepresented in red (4 5) gray (= 1) and blue (= 0) with the proportion of sites within each class represented by the color width Thickerbranches have been classified as undergoing episodic diversifying selection at corrected P 0001 (thickest branches) P 001 (medium thickness) andP 005 (thin branches) (B) Phylogenetic location of mutations involved in diversifying selection episodes during the evolution of HMGN genesBranches in red account for higher numbers of nonsynonymous mutations whereas branches in blue indicate higher numbers of synonymousmutations and branches in green represent cases with equal numbers of nonsynonymous and synonymous mutations Codon 49 is located withinthe highly conserved NDB region

128

Gonzalez-Romero et al doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

sequence alignments were conducted on the basis of thetranslated amino acid sequences and edited for potentialerrors using the BIOEDIT (Hall 1999) and ClustalW(Thompson et al 1994) programs The alignment of the com-plete set of sequences consisted of 1395 nt positions corre-sponding to 465 amino acid sites (supplementary fig S1Supplementary Material online) The boundaries of N-termi-nal (including the NBD) and acidic C-terminal regions ofHMGN proteins (containing the RD) were established onthe basis of the information available in literature as followsHMGN1 N-terminus nucleotide positions 1ndash153 C-terminuspositions 154ndash342 (Ding et al 1997) HMGN2 N-terminuspositions 1ndash147 C-terminus positions 148ndash279 (Crippaet al 1992) HMGN3 N-terminus positions 1ndash147 C-terminuspositions 148ndash396 (West et al 2001) HMGN4 N-terminuspositions 1ndash141 C-terminus positions 142ndash273 (Birger et al2001) HMGN5 N-terminus positions 1ndash126 C-terminus po-sitions 127ndash1314 (King and Francomano 2001)

Phylogenetic Analysis of HMGNs

Molecular evolutionary analyses were performed using thecomputer program MEGA version 6 (Tamura et al 2013)except where noted Due to their smaller variance (Nei andKumar 2000) nucleotide and protein sequence divergencewas estimated using uncorrected differences (p-distancespartial deletion 95) The numbers of synonymous (pS) andnonsynonymous (pN) nucleotide differences per site werecomputed using the modified method of NeindashGojobori(Zhang et al 1998) providing the transitiontransversionratio (R) for each case and estimating standard errors byusing the bootstrap (BS) method (1000 replicates) HMGNphylogenies were reconstructed following a maximum like-lihood (ML) approach with the substitution models that bestfit the analyzed sequences being JTT (Jones et al 1992) andTN93 (Tamura and Nei 1993) including gamma-distributedvariation across sites for protein and nucleotide sequencesrespectively Additional HMGN phylogenies were inferred in

mammals (the only group in which all five HMGN lineagesare represented) including Human (Homo sapiens) chim-panzee (Pan troglodytes) orangutan (Pongo abelii) rhesusmacaque (Macaca mulatta) mouse (Mus musculus) rat(Rattus norvegicus) and cow (Bos taurus) The reliability ofthe reconstructed topologies was contrasted in each case bynonparametric BS (1000 replicates) and further examined bybayesian analysis using the program BEAST version 17(Drummond et al 2012) producing posterior probabilitiesThree independent Markov chain Monte Carlo runs of10000000 generations each were performed to generate pos-terior probabilities sampling tree topologies every 1000 gen-erations to ensure the independence of successive trees anddiscarding the first 1000 trees of each run as burn-in Treeswere rooted with the human HMGA1a a HMG protein func-tionally unrelated to HMGNs (Friedmann et al 1993)

Molecular Evolution and Selection Analyses

The footprint of selection on HMGN genes was studied usingtwo major approaches First descriptive analyses of nucleo-tide variation and the mode of evolution displayed byHMGNs were carried out Accordingly the numbers of syn-onymous (pS) and nonsynonymous (pN) nucleotide differ-ences per site were compared using codon-based Z-tests forselection setting the null hypothesis as H0 pS = pN and thealternative hypothesis as H1 pS4 pN (Nei and Kumar 2000)Additionally the amount of codon usage bias and the pres-ence of global and local molecular clocks were investigatedusing the programs DnaSP version 5 (Librado and Rozas 2009)and HyPhy (Pond et al 2005) respectively Finally the rates ofevolution of different HMGN lineages were estimated by cor-relating pairwise protein divergences between pairs of taxawith their corresponding divergence as defined by theTimeTree database (Hedges et al 2006) (see supplementarytable S2 Supplementary Material online) Regression analyseswere implemented using the program STATGRAPHICS Plusversion 51 (Warrenton VA)

Second the presence of lineages displaying evidence ofdiversifying (adaptive) selection episodes (4 1) was exam-ined across HMGN evolution by using the branch-site RELmodel (Pond and Frost 2005) To this end a total of 444codon positions were examined using an ML phylogenythat was reconstructed using HMGN nucleotide coding re-gions as a reference (in this instance the best-fit model ofevolution was defined as TN93 + G) no prior assumptionsabout which lineages have been subject to diversifying selec-tion were made The proportion of sites inferred to be evolv-ing under diversifying selection at each branch wereestimated using likelihood ratio tests resulting in a P valuefor episodic selection The strength of selection was parti-tioned for descriptive purposes into three categories(4 5 = 1 = 0) using three different significancelevels (Plt 0001 Plt 001 and Plt 005) to assess the ob-tained results Additionally the presence of selection at indi-vidual sites was assessed by using different codon-based MLmethods including SLAC FEL REL FUBAR and MEME withthis latter one modeling variable (dNdS) across lineages at

Table 3 Codon Positions Potentially Subject to Selection duringHMGN Evolution in Mammalsa

Codon SLAC(P value)

FEL(P value)

REL(Bayesfactor)

FUBAR(posterior

probability)

MEME(P value)

49 0687 0783 0002 0446 0009

53 0491 0552 0006 0563 0038

97 0722 0786 0003 0316 0006

128 0000 0086 6582 0767 0103

135 0000 0096 7228 0750 0076

278 0000 0195 77197 0950 0225

185 0000 0111 109453 0920 0146

196 0000 0451 72258 0821 0448

363 0000 0082 52196 0942 0031

376 0000 0460 53381 0801 0423

431 0000 0233 57856 0853 0025

433 0000 0140 8671 0802 0037

aPositions subject to selection () as identified by the codon-based ML methodsused to estimate at different positions

129

Long-Term Evolution of HMGN Proteins doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

an individual site (Murrell et al 2012) A total of seven codonssubject to significant episodes of diversifying selection(Plt 005) were detected using MEME and analyzed in thecontext of the HMGN phylogeny providing information oninternal branches accumulating higher numbers of nonsyn-onymous mutations All analyses in this section were carriedout using the HyPhy program (Pond et al 2005) and theDatamonkey webserver (Poon et al 2009 Delport et al 2010)

Supplementary MaterialSupplementary tables S1 and S2 and figures S1 and S2 areavailable at Molecular Biology and Evolution online (httpwwwmbeoxfordjournalsorg)

Acknowledgments

This work was supported by a Canadian Institutes of HealthResearch (CIHR) grant (MOP-97878) to JA RG-R is the re-cipient of a postdoctoral fellowship from the Spanish Ministryof Education JME-L has been supported by a start-up grantfrom the College of Arts and Sciences at Florida InternationalUniversity (CAS-FIU)

ReferencesBelova GI Postnikov YV Furusawa T Birger Y Bustin M 2008

Chromosomal protein HMGN1 enhances the heat shock-inducedremodeling of Hsp70 chromatin J Biol Chem 2838080ndash8088

Bergel M Herrera JE Thatcher BJ Prymakowska-Bosak M Vassilev ANakatani Y Martin B Bustin M 2000 Acetylation of novel sites inthe nucleosomal binding domain of chromosomal protein HMG-14by p300 alters its interaction with nucleosomes J Biol Chem 27511514ndash11520

Bianchi ME Agresti A 2005 HMG proteins dynamic players in generegulation and differentiation Curr Opin Genet Dev 15496ndash506

Birger Y Catez F Furusawa T Lim JH Prymakowska-Bosak M West KLPostnikov YV Haines DC Bustin M 2005 Increased tumorigenicityand sensitivity to ionizing radiation upon loss of chromosomal pro-tein HMGN1 Cancer Res 656711ndash6718

Birger Y Ito Y West KL Landsman D Bustin M 2001 HMGN4 a newlydiscovered nucleosome-binding protein encoded by an intronlessgene DNA Cell Biol 20257ndash264

Browne DL Dodgson JB 1993 The gene encoding chicken chromo-somal protein HMG-14a is transcribed into multiple mRNAsGene 124199ndash206

Bustin M 1999 Regulation of DNA-dependent activities by the func-tional motifs of the high-mobility-group chromosomal proteins MolCell Biol 195237ndash5246

Bustin M 2001a Chromatin unfolding and activation by HMGN()chromosomal proteins Trends Biochem Sci 26431ndash437

Bustin M 2001b Revised nomenclature for high mobility group (HMG)chromosomal proteins Trends Biochem Sci 26152ndash153

Bustin M Reeves R 1996 High-mobility-group chromosomal proteinsarchitectural components that facilitate chromatin function ProgNucleic Acid Res Mol Biol 5435ndash100

Catez F Brown DT Misteli T Bustin M 2002 Competition betweenhistone H1 and HMGN proteins for chromatin binding sites EMBORep 3760ndash766

Chen P Wang XL Ma ZS Xu Z Jia B Ren J Hu YX Zhang QH Ma TGYan BD et al 2012 Knockdown of HMGN5 expression by RNAinterference induces cell cycle arrest in human lung cancer cellsAsian Pac J Cancer Prev 133223ndash3228

Crippa MP Alfonso PJ Bustin M 1992 Nucleosome core binding regionof chromosomal protein HMG-17 acts as an independent functionaldomain J Mol Biol 228442ndash449

Delport W Poon AF Frost SD Kosakovsky Pond SL 2010 Datamonkey2010 a suite of phylogenetic analysis tools for evolutionary biologyBioinformatics 262455ndash2457

Ding HF Bustin M Hansen U 1997 Alleviation of histone H1-mediatedtranscriptional repression and chromatin compaction by the acidicactivation region in chromosomal protein HMG-14 Mol Cell Biol 175843ndash5855

Drummond AJ Suchard MA Xie D Rambaut A 2012 Bayesian phylo-genetics with BEAUti and the BEAST 17 Mol Biol Evol 291969ndash1973

Eirın-Lopez JM Ishibashi T Ausio J 2008 H2ABbd a quickly evolvinghypervariable mammalian histone that destabilizes nucleosomes inan acetylation-independent way FASEB J 22316ndash326

Finn RM Browne K Hodgson KC Ausio J 2008 sNASP a histoneH1-specific eukaryotic chaperone dimer that facilitates chromatinassembly Biophys J 951314ndash1325

Friedmann M Holth LT Zoghbi HY Reeves R 1993 Organization in-ducible-expression and chromosome localization of the humanHMG-I(Y) nonhistone protein gene Nucleic Acids Res 214259ndash4267

Furusawa T Cherukuri S 2010 Developmental function of HMGN pro-teins Biochim Biophys Acta 179969ndash73

Gerlitz G Hock R Ueda T Bustin M 2009 The dynamics of HMGprotein-chromatin interactions in living cells Biochem Cell Biol 87127ndash137

Goodwin GH Walker JM Johns EW 1978 Studies on the degradation ofhigh mobility group non-histone chromosomal proteins BiochimBiophys Acta 519233ndash242

Green J Ikram M Vyas J Patel N Proby CM Ghali L Leigh IM OrsquoTooleEA Storey A 2006 Overexpression of the Axl tyrosine kinase recep-tor in cutaneous SCC-derived cell lines and tumours Br J Cancer 941446ndash1451

Hall TA 1999 BioEdit a user friendly biological sequence alignmenteditor and analysis program for Windows 9598NT Nucleic AcidsSymp Ser 4195ndash98

Hedges SB Dudley J Kumar S 2006 TimeTree a public knowledge-base ofdivergence times among organisms Bioinformatics 222971ndash2972

Hock R Scheer U Bustin M 1998 Chromosomal proteins HMG-14 andHMG-17 are released from mitotic chromosomes and importedinto the nucleus by active transport J Cell Biol 1431427ndash1436

Hock R Wilde F Scheer U Bustin M 1998 Dynamic relocation ofchromosomal protein HMG-17 in the nucleus is dependent ontranscriptional activity Embo J 176992ndash7001

Ishibashi T Li A Eirın-Lopez JM Zhao M Missiaen K Abbott DWMeistrich ML Hendzel MJ Ausio J 2010 H2ABbd an X-chromo-some-encoded histone involved in mammalian spermiogenesisNucleic Acids Res 381780ndash1789

Ito Y Bustin M 2002 Immunohistochemical localization of the nucle-osome-binding protein HMGN3 in mouse brain J HistochemCytochem 501273ndash1275

Ji SQ Yao L Zhang XY Li XS Zhou LQ 2012 Knockdown of the nu-cleosome binding protein 1 inhibits the growth and invasion of clearcell renal cell carcinoma cells in vitro and in vivo J Exp Clin CancerRes 3122

Jiang N Zhou LQ Zhang XY 2010 Downregulation of the nucleosome-binding protein 1 (NSBP1) gene can inhibit the in vitro and in vivoproliferation of prostate cancer cells Asian J Androl 12709ndash717

Johns EW 1982 The HMG chromosomal proteins New York AcademicPress

Johnson KR Cook SA Bustin M Davisson MT 1992 Genetic mappingof the murine gene and 14 related sequences encoding chromo-somal protein HMG-14 Mamm Genome 3625ndash632

Johnson KR Cook SA Ward-Bailey P Bustin M Davisson MT 1993Identification and genetic mapping of the murine gene and 20related sequences encoding chromosomal protein HMG-17Mamm Genome 483ndash89

Jones DT Taylor WR Thornton JM 1992 The rapid generation of mu-tation data matrices from protein sequences Comput Appl Biosci 8275ndash282

130

Gonzalez-Romero et al doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from

Kasinsky HE Lewis JD Dacks JB Ausio J 2001 Origin of H1 linker his-tones FASEB J 1534ndash42

Kato H van Ingen H Zhou BR Feng H Bustin M Kay LE Bai Y 2011Architecture of the high mobility group nucleosomal protein 2-nu-cleosome complex as revealed by methyl-based NMR Proc NatlAcad Sci U S A 10812283ndash12288

Kim YC Gerlitz G Furusawa T Catez F Nussenzweig A Oh KS KraemerKH Shiloh Y Bustin M 2009 Activation of ATM depends on chro-matin interactions occurring before induction of DNA damage NatCell Biol 1192ndash96

King LM Francomano CA 2001 Characterization of a human geneencoding nucleosomal binding protein NSBP1 Genomics 71163ndash173

Kosakovsky Pond SL Frost SD 2005 Not so different after all a com-parison of methods for detecting amino acid sites under selectionMol Biol Evol 221208ndash1222

Kuehl L Salmond B Tran L 1984 Concentrations of high-mobility-group proteins in the nucleus and cytoplasm of several rat tissuesJ Cell Biol 99648ndash654

Kugler JE Deng T Bustin M 2012 The HMGN family of chromatin-binding proteins dynamic modulators of epigenetic processesBiochim Biophys Acta 1819652ndash656

Laemmli UK Johnson RA 1973 Maturation of the head of bacterio-phage T4 II Head-related aberrant tau-particles J Mol Biol 80601ndash611

Li DQ Hou YF Wu J Chen Y Lu JS Di GH Ou ZL Shen ZZ Ding J ShaoZM 2006 Gene expression profile analysis of an isogenic tumourmetastasis model reveals a functional role for oncogene AF1Q inbreast cancer metastasis Eur J Cancer 423274ndash3286

Librado P Rozas J 2009 DnaSP v5 a software for comprehensive analysisof DNA polymorphism data Bioinformatics 251451ndash1452

Lim JH Bustin M Ogryzko VV Postnikov YV 2002 Metastable macro-molecular complexes containing high mobility group nucleosome-binding chromosomal proteins in HeLa nuclei J Biol Chem 27720774ndash20782

Lim JH Catez F Birger Y Postnikov YV Bustin M 2004 Preparation andfunctional analysis of HMGN proteins Methods Enzymol 375323ndash342

Luger K Mader AW Richmond RK Sargent DF Richmond TJ 1997Crystal structure of the nucleosome core particle at 28 A resolutionNature 389251ndash260

Malicet C Rochman M Postnikov Y Bustin M 2011 Distinct propertiesof human HMGN5 reveal a rapidly evolving but functionally con-served nucleosome binding protein Mol Cell Biol 312742ndash2755

Murrell B Wertheim JO Moola S Weighill T Scheffler K KosakovskyPond SL 2012 Detecting individual sites subject to episodic diver-sifying selection PLoS Genet 8e1002764

Nei M Kumar S 2000 Molecular evolution and phylogenetics NewYork Oxford University Press

Pogna EA Clayton AL Mahadevan LC 2010 Signalling to chromatinthrough post-translational modifications of HMGN BiochimBiophys Acta 179993ndash100

Pond SL Frost SD 2005 A genetic algorithm approach to detectinglineage-specific variation in selection pressure Mol Biol Evol 22478ndash485

Pond SL Frost SD Muse SV 2005 HyPhy hypothesis testing usingphylogenies Bioinformatics 21676ndash679

Poon AF Frost SD Pond SL 2009 Detecting signatures of selection fromDNA sequences using Datamonkey Methods Mol Biol 537163ndash183

Popescu N Landsman D Bustin M 1990 Mapping the human genecoding for chromosomal protein HMG-17 Hum Genet 85376ndash378

Postnikov Y Bustin M 2010 Regulation of chromatin structure andfunction by HMGN proteins Biochim Biophys Acta 179962ndash68

Postnikov YV Herrera JE Hock R Scheer U Bustin M 1997 Clusters ofnucleosomes containing chromosomal protein HMG-17 in chroma-tin J Mol Biol 274454ndash465

Postnikov YV Trieschmann L Rickers A Bustin M 1995 Homodimersof chromosomal proteins HMG-14 and HMG-17 in nucleosomecores J Mol Biol 252423ndash432

Prymakowska-Bosak M Misteli T Herrera JE Shirakawa H Birger YGarfield S Bustin M 2001 Mitotic phosphorylation prevents thebinding of HMGN proteins to chromatin Mol Cell Biol 215169ndash5178

Qu J Yan R Chen J Xu T Zhou J Wang M Chen C Yan Y Lu Y 2011HMGN5a potential oncogene in gliomas J Neurooncol 104729ndash736

Rochman M Malicet C Bustin M 2010 HMGN5NSBP1 a newmember of the HMGN protein family that affects chromatin struc-ture and function Biochim Biophys Acta 179986ndash92

Rochman M Postnikov Y Correll S Malicet C Wincovitch S KarpovaTS McNally JG Wu X Bubunenko NA Grigoryev S et al 2009 Theinteraction of NSBP1HMGN5 with nucleosomes in euchromatincounteracts linker histone-mediated chromatin compaction andmodulates transcription Mol Cell 35642ndash656

Shirakawa H Landsman D Postnikov YV Bustin M 2000 NBP-45 anovel nucleosomal binding protein with a tissue-specific and devel-opmentally regulated expression J Biol Chem 2756368ndash6374

Srikantha T Landsman D Bustin M 1987 Retropseudogenes for humanchromosomal protein HMG-17 J Mol Biol 197405ndash413

Strichman-Almashanu LZ Bustin M Landsman D 2003 Retroposedcopies of the HMG genes a window to genome dynamicsGenome Res 13800ndash812

Tamura K Nei M 1993 Estimation of the number of nucleotide sub-stitutions in the control region of mitochondrial DNA in humansand chimpanzees Mol Biol Evol 10512ndash526

Tamura K Stecher G Peterson D Filipski A Kumar S 2013 MEGA6molecular evolutionary genetics analysis version 60 Mol Biol Evol302725ndash2729

Thompson JD Higgins DG Gibson TJ 1994 CLUSTAL W improving thesensitivity of progressive multiple sequence alignments through se-quence weighting position specific gap penalties and weight matrixchoice Nucl Acids Res 224673ndash4680

Trieschmann L Martin B Bustin M 1998 The chromatin unfoldingdomain of chromosomal protein HMG-14 targets the N-terminaltail of histone H3 in nucleosomes Proc Natl Acad Sci U S A 955468ndash5473

Ueda T Catez F Gerlitz G Bustin M 2008 Delineation of the proteinmodule that anchors HMGN proteins to nucleosomes in the chro-matin of living cells Mol Cell Biol 282872ndash2883

Ueda T Furusawa T Kurahashi T Tessarollo L Bustin M 2009 Thenucleosome binding protein HMGN3 modulates the transcriptionprofile of pancreatic beta cells and affects insulin secretion Mol CellBiol 295264ndash5276

Ueda T Postnikov YV Bustin M 2006 Distinct domains in high mobilitygroup N variants modulate specific chromatin modifications J BiolChem 28110182ndash10187

Vestner B Bustin M Gruss C 1998 Stimulation of replication efficiencyof a chromatin template by chromosomal protein HMG-17 J BiolChem 2739409ndash9414

West KL Castellini MA Duncan MK Bustin M 2004 Chromosomalproteins HMGN3a and HMGN3b regulate the expression of glycinetransporter 1 Mol Cell Biol 243747ndash3756

West KL Ito Y Birger Y Postnikov Y Shirakawa H Bustin M 2001HMGN3a and HMGN3b two protein isoforms with a tissue-specificexpression pattern expand the cellular repertoire of nucleosome-binding proteins J Biol Chem 27625959ndash25969

Wu J Kim S Kwak MS Jeong JB Min HJ Yoon HG Ahn JH Shin JS 2014High mobility group nucleosomal binding domain 2 (HMGN2)SUMOylation by the SUMO E3 ligase PIAS1 decreases the bindingaffinity to nucleosome core particles J Biol Chem 28920000ndash20011

Zhang J Rosenberg HF Nei M 1998 Positive Darwinian selection aftergene duplication in primate ribonuclease genes Proc Natl Acad SciU S A 953708ndash3713

Zhou BR Feng H Kato H Dai L Yang Y Zhou Y Bai Y 2013 Structuralinsights into the histone H1-nucleosome complex Proc Natl AcadSci U S A 11019390ndash19395

Zhu N Hansen U 2010 Transcriptional regulation by HMGN proteinsBiochim Biophys Acta 179974ndash79

131

Long-Term Evolution of HMGN Proteins doi101093molbevmsu280 MBE at Florida International U

niversity on Decem

ber 22 2014httpm

beoxfordjournalsorgD

ownloaded from