Báo cáo khoa học: Serine-arginine protein kinases: a small protein kinase family with a large cellular presence potx

The SRPK1 and Dsk1 nucleotide sequencing identified a domain interrupting the kinase catalytic site into two structural entities, Keywords LBR; metabolic signalling; nuclear envelope; p53

Serine-arginine protein kinases: a small protein kinase

family with a large cellular presence

Thomas Giannakouros1, Eleni Nikolakaki1, Ilias Mylonis2 and Eleni Georgatsou2

1 Laboratory of Biochemistry, Department of Chemistry, Aristotle University of Thessaloniki, Greece

2 Laboratory of Biochemistry, Department of Medicine, School of Health Sciences, University of Thessaly, Larissa, Greece

History of the discovery of the

serine-arginine protein kinase

(SPRK) family

The first serine-arginine (SR) protein kinase to be

puri-fied and characterized was named SRPK1, for

SR-pro-tein-specific kinase 1 [1,2] It was isolated during a

search for the activity that phosphorylates SR splicing

factors (also named SR proteins) during mitosis

SRPK1 was shown to phosphorylate SR proteins in a

cell-cycle regulated manner, to affect SR protein

locali-zation and to inhibit splicing when added in large

quantities to a cell-free splicing assay [1,2] The SRPK1 cDNA was cloned, revealing that the Schizo-saccharomyces pombe SRPK1 orthologue, Dsk1, had already been cloned and partially characterized as a kinase with cell cycle-dependent phosphorylation and subcellular localization [3] The SRPK1 and Dsk1 nucleotide sequencing identified a domain interrupting the kinase catalytic site into two structural entities,

Keywords

LBR; metabolic signalling; nuclear envelope;

p53; PGC-1; protamine; spermatogenesis;

splicing; SR protein; SRPK

Correspondence

E Georgatsou, Laboratory of Biochemistry,

Department of Medicine, School of Health

Sciences, University of Thessaly, Biopolis,

41110 Larissa, Greece

Fax: +30 2410 685545

Tel: +30 2410 685581

E-mail: egeorgat@med.uth.gr

(Received 7 July 2010, accepted 26 October

2010)

doi:10.1111/j.1742-4658.2010.07987.x

Serine-arginine protein kinases (SPRKs) constitute a relatively novel subfamily of serine-threonine kinases that specifically phosphorylate serine residues residing in serine-arginine⁄ arginine-serine dipeptide motifs Fifteen years of research subsequent to the purification and cloning of human SRPK1 as a SR splicing factor-phosphorylating protein have lead to the accumulation of information on the function and regulation of the different members of this family, as well as on the genomic organization of SRPK genes in several organisms Originally considered to be devoted to constitu-tive and alternaconstitu-tive mRNA splicing, SRPKs are now known to expand their influence to additional steps of mRNA maturation, as well as to other cellular activities, such as chromatin reorganization in somatic and sperm cells, cell cycle and p53 regulation, and metabolic signalling Similarly, SRPKs were considered to be constitutively active kinases, although several modes of regulation of their function have been demonstrated, implying an elaborate cellular control of their activity Finally, SRPK gene sequence information from bioinformatics data reveals that SRPK gene homologs exist either in single or multiple copies in every single eukaryotic organism tested, emphasizing the importance of SRPK protein function for cellular life

Abbreviations

CDK, cyclin dependent kinase; Clk, CDK-like kinase; CK2, casein kinase 2; FOXO1, forkhead box protein O1; HBV, hepatitis B virus; HP1, heterochromatin protein 1; Hsp, heat shock protein; LBR, lamin B receptor; NRF-1, nuclear respiratory factor-1; PGC-1, peroxisome proliferator activated receptor c coactivator-1; RS, arginine-serine; SAFB, scaffold attachment factor B; SR, serine-arginine;

SRPK, serine-arginine protein kinase.

hence called ‘the spacer domain’, which is

characteris-tic of the SR protein kinase family [3,4] Subsequently,

Dsk1 was also shown to be a SR protein kinase

phos-phorylating and regulating the function of SR proteins

[5–7]

In 1998, the cloning of SRPK2 was reported almost

simultaneously in mouse (together with mSRPK1) [8]

and man [9] SRPK2 was found to be a SR-specific

protein kinase highly homologous to SRPK1 It is

structurally differentiated from SRPK1 by a

proline-rich tract at its N-terminus and an acidic region in its

spacer domain However, none of these elements had

been related to a particular SRPK2-specific function

until recently, when a study showed that sequences

residing in the acidic domain of SRPK2 specifically

interact with the pro-apoptotic arginine-serine (RS)

domain-containing protein acinus [10] Moreover, the

mRNA of SRPK2 was shown to have a different

and more limited tissue distribution than SRPK1

mRNA [9]

The cloning and characterization of the budding

yeast Saccharomyces cerevisiae SR protein kinase Sky1

in 1999 came not only unexpectedly, but also as a

challenge because the prevalence of uninterrupted

genes and the lack of alternative splicing in this

organ-ism would not, at that time, account for a SR

protein-specific kinase [11] Sky1 was indeed shown to have

the structural and functional characteristics of a SR

protein kinase because it could not only phosphorylate

mammalian SR splicing factors in vivo [12], but also

native RS domain-containing S cerevisiae proteins,

such as Npl3p, which is involved in mRNA export

[13] This observation led to the discovery of the

involvement of SRPKs in the regulation of additional

steps of mRNA maturation and added to the current

image of coupled transcript processing from the

tran-scription to translation steps

The cloning of SPK-1, the unique homolog of

SRPK1 in Caenorhabditis elegans, revealed an essential

function of the kinase in the germline development

and embryogenesis of this organism [14] The

underly-ing mechanism for this function has not yet been

eluci-dated, although the finding that human SRPK1 is

highly expressed in testis and phosphorylates

prot-amine 1, a highly basic protein replacing histones

dur-ing spermiogenesis, could be related with the

observations in C elegans [15]

SRPK1a, a product of the SRPK1 gene produced by

alternative splicing, that retains an additional domain

corresponding to an intron at its N-terminal region,

was reported in 2001 [16] Interestingly, this domain is

rich in proline residues reminiscent of the proline-rich

SRPK2-specific track Additionally, the SRPK1a

N-terminus was found to interact with the nuclear matrix protein scaffold attachment factor (SAFB) B1, and it was subsequently shown that SAFB proteins are inhibitors of SRPK1 and SRPK1a activity, function-ally differentiating between the two kinases and further implicating SRPKs in subnuclear organization and chromatin regulation [17]

Mouse SRPK3 was discovered in 2005, having been identified in a screen for target genes of the transcrip-tion factor myocyte enhancer factor 2 [18] SRPK3 is expressed in a tissue-specific fashion in the heart and skeletal muscle and is required for normal muscle growth and homeostasis because Srpk3-null mice suffer from centronuclear myopathy [18] It has not been confirmed, however, whether SR kinase activity is required for these phenotypes and, if so, what sub-strates are affected The existence of the orthologue of mSRPK3 in humans has been postulated in an analysis

of human chromosomal DNA methylation, although

no studies are available for its expression or function However, the cDNA of the porcine SRPK3 has been cloned and shown to have a very limited and tissue specific expression in muscular tissue [19]

Although Drosophila harbors several SRPK homo-logs, only two very recent studies refer to Srpk79D (as named by both groups), which is considered to be a product of the CG11489 gene in the Drosophila genome [20,21] It is interesting that Srpk79D displays tissue specific expression in neuronal tissue and is implicated in the development and growth of synaptic connections throughout the nervous system

Finally, some SR protein kinases of lower organisms have also been cloned, adding to the picture of SRPK function and importance TcSRPK, the SR protein kinase of a protozoan, the parasite Trypanosoma cruzi, which displays trans- and cis-splicing and was cloned and characterized in 2003, functions as a bona fide SR protein kinase, indicating that the general control of eukaryotic mRNA processing evolved early during evolution [22] More recently, PSRPK, the SR protein kinase of Physarum polycephalum, a slime mold, has been cloned and characterized, especially with respect

to its subcellular localization properties [23]

Evolution of the SRPK gene family

A simple search for genes (using the keyword ‘SRPK*’

at http://www.ncbi.nlm.nih.gov) returns approximately

90 hits for SRPK genes or putative SRPK genes in dif-ferent eukaryotic organisms Some of these sequence entries have not yet been subjected to a final NCBI review and overlaps may exist between them that should be thoroughly examined In our preliminary

research, however, we have made some interesting

observations that we note below to emphasize the

sig-nificance of evolutionary-oriented studies on the

sequences of the SRPK gene products

The first intriguing observation is that the SRPK gene

copy number of an organism does not appear to directly

relate to its evolutionary scale For example, there

exist fungi with one, two or even up to nine SRPK

genes [S cerevisiae and S pombe with one gene (Sky1

and Dsk1, respectively); Candida albicans with two

(QSAA48 and QS9Q27); Aspergilus niger with nine

(A2QAE4, A2QB94, A2QC46, A5AB23, A2QWQ2,

A2QX01, A2QX98, A2R2M0 and A2RSV1)]; plants

with three genes (Ricinus communis; B9SRL4, B9S6V7

and B9SNS8); and insects with two (Culex

quinquefasci-atus; BOWGI3 and BOWRV4) or three genes

[Drosoph-ila melanogaster; CG8174, CG8565 and CG11489

(CG9085)], whereas mammals (rat, mouse, human, etc.)

have three genes Additionally, as we have experienced

from our own research and as mentioned in the two studies

concerning Srpk79D in Drosophila melanogaster [20,21],

there is no prominent one-to-one correspondence

between the sequences of SRPK genes of evolutionary

remote species The emerging image is reminiscent of

independent SRPK gene duplications that have taken

place at several time points during evolution in different

species Accordingly, it is suggested that the SRPK

genes are subjected to an evolutionary drive that

demands multiple SRPK gene copies in almost each new

emerging species One may observe evidence of the

errors of the evolutionary ‘trial and error’ process

oper-ating through new SRPK genes: pseudogenes exist for

both SRPK1 and SRPK2 in the human genome and

also for SRPK1 in the mouse [24] Other loci identified

by sequencing might also correspond to pseudogenes

The second observation concerns the ‘spacer region’

of the SRPK proteins This sequence is SRPK

family-specific (as a serine⁄ threonine kinase subfamily) and

each family member harbors its own unique spacer It

should be noted that the different ‘spacer regions’, in

addition to being very different with respect to their

primary sequence, are very diverse in length, and

pos-sibly function too, as indicated by the data presented

further below Consequently, it is not unexpected that,

for all the SRPK sequences we randomly examined

from different kingdoms, the whole spacer sequence

resides on a separate exon, suggesting that this domain

may have evolved independently

Another domain of interest that remains relatively

unexamined from an evolutionary point of view is the

N-terminal domain of the kinases It is highly specific

between the SRPK family members, and few functions

have been attributed to it It may be important to note

that it is this particular region of the mRNA that fre-quently swaps and alternates in the splicing phenom-ena that are beginning to be revealed in SRPK transcripts [16,20,21]

Finally, it is interesting to note that SRPK2 contains

a minor class of introns [25] Because the minor class

of introns is often associated with many important genes that are evolutionarily conserved, it is likely that SRPK2 is evolving and regulated by a distinct mecha-nism from SRPK1

Function of the SRPKs

As already noted, SRPKs phosphorylate their substrates at serine residues located in regions rich in arginine⁄ serine dipeptides, known as RS domains The definition of a ‘typical’ RS domain is somewhat arbi-trary and SRPKs have been shown to be able to phos-phorylate scattered RS dipeptides if they conform to certain limitations [26–29] The specificity of these enzymes is remarkable because mutations of Ser to Thr or Arg to Lys in the RS domain completely abro-gate phosphorylation [2,26]

In the list of the RS domain-containing proteins, the

SR proteins prevail, either as the originally identified

‘classical’ SR proteins invariably containing an RNA recognition motif or as ‘SR-like’ or ‘SR-related’ pro-teins also containing RNA binding domains (RNA recognition motif or other) Most of the SR splicing factors have been experimentally shown to be SRPK substrates in vitro and in vivo and it is to be expected that every SR protein could potentially be a SRPK substrate under particular cellular conditions Yet a recent study suggests that the human genome encodes for more than 100 RS domain-containing proteins [30], indicating that SRPKs may regulate diverse cellular functions through phosphorylation of many of these potential substrates Below, we review the SRPK impact on mRNA maturation and discuss the regula-tory paradigms that have been characterized to a reasonable extent, including the replacement of hist-ones by the arginine-rich protamines during spermio-genesis, the role of SPRKs in cell cycle progression and chromatin reorganization, and the function of SRPKs in the regulation of peroxisome proliferator activated receptor c coactivator (PGC)-1a in metabolic signaling

SRPKs and mRNA maturation The involvement of SRPKs in the regulation of mRNA splicing was expected because SRPK1 was iso-lated as a SR splicing factor-phosphorylating kinase

and the phosphorylation of SR proteins had been

shown to be a prerequisite for spliceosome assembly

and splicing in general [31–33] However, the exact

contribution of the SRPKs in the different steps of

mRNA maturation is not completely clarified (even up

to this date) for several reasons First, the subcellular

localization of SRPKs is cytoplasmic as well as

nuclear, implying a more complex function for these

kinases than the phosphorylation of only

cytoplasmic-or only nuclear-localized SR splicing factcytoplasmic-ors SRPKs

undoubtedly phosphorylate both, although under

con-ditions that are strictly controlled Second, SR protein

phosphorylation in the nucleus also takes place as a

result of other families of kinases The cyclin

depen-dent kinase (CDK)-like kinases (Clk) also

phosphory-late RS domains but have a much broader specificity

[26,34] Topoisomerase I has also been shown to

phosphorylate SR proteins [35] but its role in SR

protein function remains unexplored and, finally, and

also relatively recently, Akt kinases have been shown

to affect splicing by targeting RS domains [36] Third,

the specific functions of the various SRPKs

discov-ered in different organisms are just beginning to be

addressed

As already mentioned, concomitant with its purifica-tion, SRPK1 was shown to inhibit splicing in vitro when present in large quantities and to disassemble nuclear speckles when added in permeabilized cells [1] This and other in vitro experiments have implicated SRPKs in the phosphorylation of SR splicing factors and the regulation of splicing [2,9,37], although the first study to definitively attribute a role of a SRPK on

SR protein function in vivo was carried out by Yeakley

et al [12], which showed that when the unique SRPK

of S cerevisiae (Sky1) is deleted, the interaction of SR proteins is prevented, and they are incapable of trans-locating into the nucleus Importantly, that study, which used mammalian splicing factors, showed for the first time that SRPK-mediated phosphorylation plays an important role in SR protein nuclear import and that not all SR splicing factors are affected identi-cally Sequential studies with Sky1 and its SR-like pro-tein substrate Nlp3p (which transports mRNAs out of the nucleus) have shown that Nlp3p needs to be phos-phorylated to release the mRNA and be re-imported into the nucleus [13,38] In humans, shuttling splicing factors such as SF2⁄ ASF are phosphorylated in the cytoplasm by SRPK1 (Fig 1) and are subsequently

Fig 1 SRPK regulation and function in mRNA maturation During interphase, SRPKs are sequestered in the cytoplasm via their spacer domain and anchoring to various protein complexes, where they phosphorylate SR proteins and facilitate their nuclear import After stress-induced, cell cycle-dependent or other signalling, SRPKs translocate to the cell nucleus where, along with other SR protein kinases (Clks), they further modify their substrates found in nuclear speckles SRPK-mediated phosphorylation influences the dissociation of SR splicing fac-tors from speckles, spliceosome assembly and splice site selection Dephosphorylation of SR splicing facfac-tors is required for splicing activity and their export to the cytoplasm In the nucleus, SRPKs can interact with nuclear matrix proteins such as SAFB Their export is the result

of an as yet unidentified mechanism Black arrows indicate molecule reactions or movements Dashed lines indicate hypothetical molecule reactions or movements.

transported to the nucleus by transportin-SR2, which

specifically interacts with phosphorylated RS domains

[39–41]

In the nucleus, when not actively involved in

tran-script processing, SR proteins reside in nuclear speckles

from which they are released when subjected to a new

round of phosphorylation Although SRPKs are able

to phosphorylate SR proteins residing in nuclear

speck-les, it is difficult to determine the exact roles of the

different kinases involved in phosphorylation of SR

splicing factors in the nucleus (Fig 1) The importance

of the function of SRPKs for the splicing reaction

in vivofirst became apparent from the genetic approach

of Dagher & Fu [42] in budding yeast, where it was

shown that Sky1 interacts with proteins that affect

3¢ splice site selection Additionally, in co-transfection

experiments, Nikolakaki et al [16] showed that both

SRPK1 and SRPK1a are capable of altering the

splic-ing of a tau minigene in a dose-dependent manner

Moreover, it is interesting that, in humans, SRPK1 has

been found to be associated with the U1-snRNP, which

is involved in 5¢ splice selection [43], whereas SRPK2

was shown to be required for the formation of the

U4⁄ U6-U5 tri-snRNP, which is involved in 3¢ splice site

selection [44] Similarly, using siRNA, Hayes et al [45]

have confirmed the role of SRPK1 in phosphorylating

SR proteins in vivo and have connected the endogenous

down-regulation of SRPK1 expression with alternative

splicing of a particular transcript Finally, Zhong et al

[46] clearly showed that when SR protein kinases enter

the nucleus (in this case under a stress signal),

phos-phorylation of SR proteins is increased, verifying the

nuclear action of SRPKs on SR splicing factors

Accordingly, a study by Jiang et al [47] showed that

when SRPK2 is phosphorylated by Akt in neuronal

cells, it enters the nucleus and is able to phosphorylate

the nonshuttling SR splicing factor SC35

As previously noted, SR proteins are implicated in a

much broader spectrum of activities that accompany

the life of an mRNA, in addition to the splicing

pro-cess To function in mRNA export, SR proteins need

to be underphosphorylated (Fig 1) On the other

hand, SF2⁄ ASF does not have to leave the nucleus to

exert its positive effect on mRNA nonsense mediated

decay [48] An intact RS domain is required for this

particular function, yet the impact of its

phosphoryla-tion state has not been addressed In addiphosphoryla-tion,

SF2⁄ ASF has been recently shown to be a

transla-tional activator of capped mRNAs in the cytoplasm

[49]; however, no report on its state of

phosphoryla-tion was included in that study The participaphosphoryla-tion of

the SRPKs in these SR protein-dependent functions

would be an interesting subject for future studies In

this respect, it should be noted that, in yeast, where the SR-like protein Npl3 was found to also affect translation (albeit by a different mechanism than SF2⁄ ASF in humans), this activity was shown to be Sky1-independent [50]

The key role played by SRPKs in mRNA processing

is particularly apparent in studies on pathological con-ditions, such as viral infection and tumor development The herpes simplex virus-1 protein ICP27 interacts with SRPK1, relocalizing it to the nucleus and affect-ing its function, resultaffect-ing in lower total host splicaffect-ing activity, and thus favoring the exit from the nucleus of the intronless viral mRNAs [51] The E1^E4 protein of human papilloma virus 1 interacts with SRPK1 and can function as a substrate for the kinase The in vivo effects of this interaction are not known, however, nor

is it known whether these putative effects would be exerted via the splicing machinery [52] A third virus found to be directly involved in SRPK function is hep-atitis B virus (HBV) In HBV-infected cells, before the encapsidation of the virus genetic material, the unique viral core protein needs to be phosphorylated by a host kinase Although there are conflicting results as to whether the kinases responsible for this phosphoryla-tion are SRPK1 and 2, there is agreement on the fact that the viral protein interacts with SRPK1 and that this interaction affects the HBV cell cycle [53,54] This

as well as other evidence suggests that SRPKs may be potential pharmaceutical targets for the control of viral infection Hence, a small molecule, isonicotinamide compound, which is a relatively selective inhibitor of SRPK1 and 2 (SPRIN340), was found to impair Sind-bis virus propagation in cultured cells, although it is only variably effective on HIV-1 propagation [55] Interest in SRPKs as pharmaceutical targets also emerged from the observation that SRPKs show increased expression in tumors of pancreas, breast and colon [45,56], as well as in acute T-cell leukemia induced by human T-cell leukemia virus-1 [57] Accordingly, cell lines derived from pancreatic, breast and colonic tumors, when disrupted for the SRPK1 gene, display diminished cell proliferation, increased apoptotic potential and augmented sensitivity to the common chemotherapeutics gemcitabine and cisplatine Evidence has been provided that the results observed are effected through the splicing machinery [45] An inverse correlation has been documented, however, between the expression of SRPK1 and cisplatin sensi-tivity in yeast and in cells of germline origin, where down-regulation of SRPK1 confers resistance to cis-platin [58,59] These tissue-specific findings again point out the intricate and fine-tuned cellular networks regu-lated by SRPK activity

SRPKs and spermiogenesis

SRPK1, SRPK1a and SRPK2 are predominantly

expressed in testis [9,15,16] Because of the numerous

maturation stages that germ cells undergo, novel gene

regulation strategies have been developed that provide

for flexible gene expression and protein function and,

among them, alternative splicing is particularly

preva-lent The overexpression of SRPKs in testis may

there-fore suggest a contributory role for these kinases in

generating altered splice patterns across the

develop-mental program of germ cells Yet the levels of SR

proteins are not elevated in testis compared to other

tissues [60,61], implying that the high levels of SRPKs

are probably not just a result of changes in the splicing

machinery

The finding that P1 protamine from various

organ-isms satisfy the substrate specificity requirements of

SRPK1, coupled with the fact that most, if not all, of

these proteins are known to be phosphoproteins

[62,63], made them attractive candidate substrates of

SRPK1 Consistent with this hypothesis, Ser10 and

Ser8 (7RSQSRSR13) were identified as the in vivo

phosphorylation sites of mono- and di-phosphorylated

human P1 protamine [62] Indeed, SRPK1 was found

to phosphorylate human P1 protamine efficiently [15]

Protamines are highly basic, arginine-rich,

low-molecu-lar weight proteins that replace histones during the

development of spermatids into spermatozoa, a process

termed spermiogenesis [63] As a result of this

exchange, the nucleosomal-type chromatin is

trans-formed into a smooth fiber and compacted into a

volume approximately 5% of that of a somatic cell

nucleus [63,64] P1 protamine is the main member of

the family and is conserved in all vertebrates, whereas

P2 protamine has been described only in some species,

including man, stallion, hamster and mouse [63]

The deposition of protamines on sperm chromatin

and the subsequent chromatin condensation are largely

controlled by phosphorylation-dephosphorylation

events Protamines are highly phosphorylated shortly

after their synthesis and before binding to DNA [65]

Phosphorylation of P2 protamine has been shown to

be essential because deletion of the

calmodulin-depen-dent protein kinase Camk4, which phosphorylates P2

protamine, impairs the deposition of P2 protamine on

sperm chromatin, resulting in defective spermiogenesis

and male sterility [66] Phosphorylation of P1

prot-amine by SRPK1 is required for the temporal

associa-tion of P1 protamine with lamin B receptor (LBR), an

inner nuclear membrane protein that also possesses a

stretch of RS dipeptides at its nucleoplasmic NH2

-terminal domain [67] It is well known that RS

domains mediate protein–protein interactions in a phosphorylation-dependent manner [68], assuming that only one of the two RS domains is phosphorylated Phosphorylation of the P1 protamine molecules in the cytoplasm on their way to the nucleus together with a lack of LBR phosphorylation is consistent with the observed predominant cytoplasmic localization of SRPK1 and the minimal RS kinase activity detected in the nucleus of germ cells [4,15]

The association of P1 protamine with the nuclear envelope probably represents an important intermedi-ate step before its deposition on sperm chromatin In this respect, Biggiogera et al [69] reported that prota-mines initially appear at the nuclear periphery, imply-ing that the nuclear envelope might play a role in the replacement of transition proteins by protamines dur-ing spermiogenesis The detachment of P1 protamine from the nuclear envelope and its binding to DNA are probably achieved through its dephosphorylation (Fig 2) Consistent with this hypothesis, protamines were found mainly dephosphorylated in mature sperm chromatin [62,63] One possibility is that the nuclear envelope functions as a ‘working platform’ where addi-tional modifications (e.g methylation) of P1 protamine take place These modifications may not only increase the affinity of P1 protamine for sperm DNA, but also may recruit specific molecules, such as heterochromatin protein 1 (HP1), which were shown to be coupled to chromatin condensation and transcriptional silencing [64,70]

A central question concerning P1 protamine is how its transportation into the nucleus is accomplished Conceivably, this may be mediated through an active transport mechanism, similar to histone H1 and transi-tion protein 2, for which importin 5 and importin 4, respectively, are known to be responsible for their translocation into the nucleus [71,72] Consistent with this hypothesis, it has been suggested that phosphory-lation of the RS domain of the splicing factor ASF⁄ SF2 by SRPK1 results in a conformational change that facilitates its interaction with the nuclear transport receptor transportin-SR2 (an importin-b family protein), thereby mediating the shuttling of this

SR protein into the nucleus through the nuclear pore complex [41] In such a case, phosphorylation of P1 protamine by cytoplasmic SRPK1 may also promote its interaction with an as yet unknown importin family member, thereby facilitating its translocation into the nucleus The release of P1 protamine from importin may be mediated through its binding to LBR at the nuclear periphery

Finally, SRPKs may have additional roles in sper-matogenesis that need to be further characterized For

example, SRPK1 was reported to mediate the uptake

of polyamines through an as yet unidentified signaling

pathway [73]

SRPKs, cell cycle progression and chromatin

reorganization

SRPKs have been characterized as cell cycle regulated

kinases [1,3] This characterization was mainly based

on the finding that SRPK1, as well as its fission yeast

homolog, Dsk1, can translocate into the nucleus at the

end of the G2 phase [3,4] In addition, SRPK1 activity,

when assayed using SC35 or ASF⁄ SF2 as substrate,

has been reported to be approximately five-fold higher

in extracts from metaphase compared to interphase

cells [1] The break-up of the speckled pattern and the

redistribution of splicing factors throughout the

cyto-plasm were initially considered as the main mitotic

functions of SRPK1 [1] In this review, we discuss data

associating SRPKs not only with additional mitotic

events, but also with other functional aspects of the

mammalian cell cycle

Regulation of chromatin binding to the nuclear

envelope

Several macromolecular complexes are assembled by

various integral proteins of the nuclear envelope that

have been proposed to function as

chromatin-anchor-age platforms [74] LBR is one of the key factors that

has been implicated in chromatin anchorage and was shown to form oligomeric stuctures at the level of the nuclear envelope [75–77] The LBR–chromatin associa-tion is probably mediated by electrostatic interacassocia-tions between the positively-charged residues of the N-termi-nal domain of LBR and the negatively-charged phos-phate groups of DNA [78] The N-terminal domain of LBR harbors a RS domain, the phosphorylation of which not only reduces the positive charges, thereby weakening the interaction with DNA, but also may result in the disassembly of the oligomeric structure of LBR [79] The joint effect of the charge reduction and the conformational change may render the phosphory-lated monomeric N-terminal domains unable to anchor the arrays of nucleosomes to the nuclear periphery It

is well known that, during mitosis, the nuclear enve-lope breaks down and chromosomes dissociate from the inner nuclear membrane Already at prophase, binding of the membranous structures to chromosomes

is weakened The RS domain of LBR is phosphoryla-ted at the beginning of mitosis by nuclear-translocaphosphoryla-ted SRPK1 and potentially by Akt and Clk kinases that may also target RS domains [26,36] Furthermore, the central mitotic kinase, cdk1, phosphorylates LBR at Ser71 [80], which is located just upstream of the RS repeats It is therefore possible that these combinato-rial phosphorylation events may result in chromosome dissociation This idea is consistent with a previous study reporting that phosphorylation of LBR by mito-tic extracts impairs chromatin association [81]

Fig 2 A model illustrating the interactions between the NH2-terminal nucleoplasmic domain of LBR and P1 protamine At the beginning of spermiogenesis, the RS domain of LBR is unphosphorylated, allowing its association with phosphorylated protamine 1 LBR may act as a docking site for the replacement of transition proteins (TP) by P1 protamine in certain chromatin layers that come close to the nuclear periphery Enzymes trapped in the inner nuclear membrane (INM) may also further modify the P1 protamine molecules, thereby facilitating their deposition on sperm chromatin The detachment of P1 protamine from the nuclear envelope and its tight binding to DNA is proposed

to occur through its dephosphorylation, whereas, at the same time, a similar dephosphorylation event may trigger the dissociation of TP from sperm chromatin.

Regulation of chromatin reorganization during

G2⁄ M phase progression

Another mode of action of SRPK1 related to its

nuclear translocation at the G2⁄ M boundary involves

chromatin reorganization A recent study

demon-strated that two SR proteins, SRp20 and ASF⁄ SF2,

are released from mitotic chromosomes, during which

the H3 tail is modified at Ser10 by the activated aurora

kinase B, and reassociate with chromatin late in M

phase [82] Hyperphosphorylation of these two SR

proteins by SRPK1 was also found to significantly

diminish their interaction with the H3 tail

Intrigu-ingly, dissociation of ASF⁄ SF2 from phosphorylated

histone H3 was required for the subsequent release of

HP1 (a key constituent of interphase hetechromatin)

from mitotic chromatin

We propose that SRPKs (and potentially members

of the Akt and Clk family of kinases) may have a key

role at the beginning of mitosis by first mediating

the detachment of peripheral heterochromatin from

the inner nuclear membrane and, subsequently, the

removal of HP1, thus leading to chromosome

conden-sation (Fig 3) An even more intriguing possibility is

that these phosphorylation events may also be

applica-ble during interphase for fine-tuning gene expression

It has been suggested by Misteli [83] that the

differen-tial regulation of gene expression might involve the

inducible ‘potentiation’ of genomic loci, with

subse-quent displacement from their chromosome territory and translocation to a transcriptionally silencing or activating microenvironment Because the coupling of chromatin domains to the nuclear envelope has been proposed to result in their transcriptional inactivation [84], and HP1 proteins are well-known constituents of

‘silent’ chromatin, the regulated nuclear translocation

of SRPKs may contribute to the re-positioning and

‘unwinding’ of specific genomic loci, thus leading to their transcriptional activation

Regulation of cyclin transcription SRPK2 has been implicated in the transcriptional regulation of two members of the cyclin family In hematopoietic cells, SRPK2 was reported to enhance cyclin A1 transcription [10], whereas, in neurons, it was shown to trigger cell cycle progression and induce apoptosis through regulation of cyclin D1 [47]

Cyclin A1 is a member of mammalian A-type cyclins and is mainly expressed in male germ cells, being essential for the passage of spermatocytes into meiosis

I [85] In addition to male germ cells, elevated levels of cyclin A1 expression have been detected in several leu-kemic cell lines as well as in hematopoietic stem cells and primitive precursors [86] Up-regulation of cyclin A1 by SRPK2 is accomplished through phosphoryla-tion of the protein acinus that contains several RS domains and its subsequent redistribution from nuclear

Fig 3 Modulation of chromatin condensation at the beginning of mitosis by SR protein kinases The combined phosphorylation of the RS domain of LBR by nuclear translocated SRPK1 and the central mitotic kinase cdk1 (and potentially by Clk and Akt kinases) results in chromo-some dissociation from the inner nuclear membrane A concomitant combined phosphorylation event [i.e phosphorylation of Ser10 of H3

by aurora B and phosphorylation of SR proteins ASF ⁄ SF2 and SRp20 (SR) by nuclear translocated SRPK1, and potentially by Clk and Akt kinases] results in HP1 release from mitotic chromatin, further facilitating chromatin condensation.

speckles to the cytoplasm [10] Acinus was originally

identified as a target of caspase-3, a cysteine protease

involved in activating chromatin condensation and

nuclear fragmentation during apoptosis; however, the

role of this protein during normal cellular growth has

not been determined [87,88] Acinus is phosphorylated

by SRPK2 at Ser422, and acinus S422D, a SRPK2

phosphorylation mimetic, was shown to enhance cyclin

A1 transcription, whereas acinus S422A, an

unphosph-orylatable mutant, was shown to block the stimulatory

effect of SRPK2 Furthermore, siRNA-mediated

down-regulation of acinus or SRPK2 resulted in cyclin

A1 repression in leukemic cells and the cells were

arrested at the G1 phase Interestingly, overexpressed

FLAG-SRPK1 was unable to associate with and

phos-phorylate recombinant acinus, indicating that the

inter-action between acinus and SRPK2 is specific [10] To

our knowledge, this is the first report of two SRPK

family members exhibiting a differential recognition

pattern towards an RS domain-containing protein

Cyclin D1 functions as a mitogenic sensor and is

one of the more frequently altered cell cycle regulators

in cancers [89] It belongs to the family of mammalian

D-type cyclins that are G1-specific Cyclin D1

associ-ates with and allosterically activassoci-ates CDK4 or CDK6,

thereby promoting restriction point progression during

the G1 phase [89] Terminally differentiated neurons

are unable to reenter the cell cycle Aberrant cell cycle

activation provokes neuronal cell death, whereas cell

cycle inhibition increases neuronal survival SRPK2

triggers cell cycle progression in neurons and induces

apoptosis through regulation of nuclear cyclin D1 [47]

According to Jang et al [47], up-regulation of cyclin

D1 in this system is not mediated through acinus

phos-phorylation but rather through inactivation of p53

More specifically, it has been proposed that SRPK2

phosphorylates and activates SC35 and, thus, it may

inactivate p53 by blocking its phosphorylation at

Ser15 [47,90] Interestingly, it has been also reported

that SC35 affects transcriptional elongation in a

gene-specific manner [91] Thus, activation of SC35 may

lead to down-regulation of specific genes, including

p53 Because p53 represses cyclin D1 expression [92],

down-regulation of p53 may also result in cyclin D1

up-regulation

In this respect, it should be noted that SRPKs have

been proposed to act as modifiers of the p53 pathway

in Drosophila (Patent WO⁄ 2002 ⁄ 099427: SRPKs as

modifiers of the p53 pathway) More specifically, a

genetic screen identified that a SRPK mutation

enhanced cell death, as induced by the expression of

p53 in the Drosophila wing Because Drosophila

con-tains more than one Srpk gene, it remains unclear

whether the regulation of p53 activity is exerted by a specific SRPK (e.g the SRPK2 homolog) and, more importantly, whether this regulation is accomplished solely through SC35

SRPKs and metabolic signaling The PGC-1 family of coactivators mediates various environmental signals, thus regulating several meta-bolic pathways in a tissue-specific manner [93] Most importantly, the PGC-1 coactivators play a critical role

in modulating glucose, lipid and energy homeostasis that become deregulated in metabolic diseases such as diabetes, obesity and cardiomyopathy The first mem-ber of the PGC-1 family, PGC-1a, was identified as a cofactor for peroxisome proliferator activated recep-tor c approximately one decade ago [94] PGC-1a activity is modulated by a large number of post-trans-lational modifications, including phosphorylation by several kinases, acetylation and deacetylation by GCN5 and silent information regulator 1, respectively,

as well as O-GlcNAcylation by O-GlcNAc transferase [95]

PGC-1a contains a RS domain that links insulin sig-nal transduction to the repression of gluconeogenesis [96] This link is mediated through phosphorylation of the RS domain that renders PGC-1a unable to coacti-vate the forkhead transcription factor forkhead box protein O1 (FOXO1), which is the main nuclear recep-tor controlling the glyconeogenic program [96–98] To date, two kinases have been implicated in the phos-phorylation of the RS domain: Akt2 and Clk2 [97,98] Akt2 phosphorylates only Ser570, which is the last ser-ine in the first repeat of RS dipeptides (RSRSRSFSR) [97], whereas Clk2 probably phosphorylates the entire

RS domain [98]

A central question that arises is whether PGC-1a is also phosphorylated by members of the SRPK family, and whether this phosphorylation can repress its tran-scriptional activity It was previously shown that SRPK1 can phosphorylate in vitro the RS domain of PGC-1a [79], although a similar phosphorylation event has not yet been shown to occur in vivo We anticipate that this type of phosphorylation may also take place

in vivo and not only by SRPK1, but also by other members of the SRPK family, to an extent propor-tional to the expression levels of SRPKs in liver Another important issue is the response to insulin Akt2 is an insulin-responsive kinase, whereas it was shown to phosphorylate Clk2 at Thr343, leading to an increase of Clk2 protein stability and therefore activity [98] Clk2 was therefore suggested to function as

an insulin-induced gluconeogenic repressor Yet Akt

kinases can also phosphorylate SRPK2 at Thr492 and

mediate its nuclear translocation [47], thus making

SRPK2 an insulin-responsive kinase as well It is still

unknown whether insulin has any effect on SRPK1

and⁄ or on SRPK1a, either directly through

phosphor-ylation by Akt kinases or through another indirect

sig-nalling mechanism In this respect, it should be noted

that SRPK1a contains two LXXLL motifs [16] that

are assumed to facilitate the interaction of different

proteins with nuclear receptors All these

phosphoryla-tion events may act in a complementary fashion

(Fig 4A), thus constituting a fine-tuning mechanism

that modulates the interaction of PGC-1a with various transcription factors and allows the expression of specific gene sets in different physiological settings PGC-1a also stimulates mitochondrial biogenesis through coactivation of nuclear respiratory factor-1 (NRF-1) [99] Indeed, PGC-1a expression in both mus-cle and fat cells activates the expression of several genes of the oxidative phosphorylation pathway, including cytochrome c oxidase subunits II and IV, and ATP synthase Although the RS domain of PGC-1a mediates its interaction with FOXO1, the 200-400 amino acid region of PGC-1a is responsible for the

Fig 4 Regulation of PGC-1a transcriptional activity by its RS region (A) Akt2, Clk2 and potentially SRPK2 phosphorylate various serine resi-dues within the RS region of PGC-1a, thus impairing its interaction with FOXO-1 to a different extent each time Akt2 phosphorylates only Ser570 (purple), SRPK2 (and possibly other SRPKs as well) may phosphorylate the serines within the three repeats of RS dipeptides (blue), whereas Clk2 probably phosphorylates the entire RS domain (red) Akt2 is activated by insulin, whereas it phosphorylates Clk2 at Thr343, leading to an increase of Clk2 protein stability and activity Akt kinases also phosphorylate SRPK2 at Thr492 and mediate its nuclear translo-cation, thus rendering SRPK2 molecules available to phosphorylate nuclear PGC-1a It is not clear yet whether there is further cross-regula-tion between SRPK, Clk and Akt kinases The stoichiometry of PGC-1a phosphorylacross-regula-tion (i.e the number of protein molecules per cell that are phosphorylated) and also the exact serines of the RS region that are modified in each molecule may constitute a fine-tuning mechanism that regulates the transcription of gluconeogenic genes and mediates PGC-1a responsiveness to insulin (B) A more permanent inactivation

of the RS domain (in muscle and fat cells) may be achieved through binding of p32 protein A central function of p32 protein is to associate with and impair the phosphorylation of RS domains p32 protein may obstruct the interaction of PGC-1a with FOXO-1 that requires the RS domain, thus allowing the available PGC-1a molecules to interact with NRF-1 and promote the transcription of specific genes involved in oxidative phosphorylation.