Báo cáo khoa học: The enzymatic activity of SR protein kinases 1 and 1a is negatively affected by interaction with scaffold attachment factors B1 and 2 pot

In this study, we show that SAFB1 inhibits the activity of both SRPK1a and SRPK1 in vitro and that its RE-rich region is redundant for the observed inhibition.. We dem-onstrate that kina

negatively affected by interaction with scaffold

attachment factors B1 and 2

Dora Tsianou1, Eleni Nikolakaki2, Alexandra Tzitzira1, Sofia Bonanou1, Thomas Giannakouros2 and Eleni Georgatsou1

1 Department of Medicine, University of Thessaly, Mezourlo, 41110 Larissa, Greece

2 Department of Chemistry, The Aristote University of Thessaloniki, Greece

Keywords

kinase activity inhibition; nuclear complex

formation; SAFB; SRPK1; SRPK1a

Correspondence

E Georgatsou, Laboratory of Biochemistry,

Department of Medicine, School of Health

Sciences, University of Thessaly, Mezourlo,

41110 Larissa, Greece

Fax: +30 241 068 5545

Tel: +30 241 068 5581

E-mail: egeorgat@med.uth.gr

Website: http://www.med.uth.gr

(Received 24 January 2009, accepted 16

July 2009)

doi:10.1111/j.1742-4658.2009.07217.x

SR protein kinases (SRPKs) phosphorylate Ser⁄ Arg dipeptide-containing proteins that play crucial roles in a broad spectrum of basic cellular processes Phosphorylation by SRPKs constitutes a major way of regulating such cellu-lar mechanisms In the past, we have shown that SRPK1a interacts with the nuclear matrix protein scaffold attachment factor B1 (SAFB1) via its unique N-terminal domain, which differentiates it from SRPK1 In this study, we show that SAFB1 inhibits the activity of both SRPK1a and SRPK1 in vitro and that its RE-rich region is redundant for the observed inhibition We dem-onstrate that kinase activity inhibition is caused by direct binding of SAFB1

to SRPK1a and SRPK1, and we also present evidence for the in vitro binding

of SAFB2 to the two kinases, albeit with different affinity Moreover, we show that both SR protein kinases can form complexes with both scaffold attach-ment factors B in living cells and that this interaction is capable of inhibiting their activity, depending on the tenacity of the complex formed Finally, we present data demonstrating that SRPK⁄ SAFB complexes are present in the nucleus of HeLa cells and that the enzymatic activity of the nuclear matrix-localized SRPK1 is repressed These results suggest a new role for SAFB proteins as regulators of SRPK activity and underline the importance of the assembly of transient intranuclear complexes in cellular regulation

Structured digital abstract

l MINT-7228149 : SRPK1 (uniprotkb: Q96SB4-2 ) phosphorylates ( MI:0217 ) Nt-LBR (uni-protkb: Q14739 ) by protein kinase assay ( MI:0424 )

l MINT-7228207 : SRPK1 (uniprotkb: Q96SB4-2 ) physically interacts ( MI:0915 ) with SAFB1C (uniprotkb: Q15424 ) by pull down ( MI:0096 )

l MINT-7228438 : SRPK1a (uniprotkb: Q96SB4-3 ) physically interacts ( MI:0915 ) with SAFB1C (uniprotkb: Q15424 ) by pull down ( MI:0096 )

l MINT-7228306 : SRPK1 (uniprotkb: Q14151 ) physically interacts ( MI:0915 ) with SAFB2C (uniprotkb: Q14151 ) by pull down ( MI:0096 )

l MINT-7228452 : SRPK1a (uniprotkb: Q96SB4-3 ) physically interacts ( MI:0915 ) with SAFB2C (uniprotkb: Q14151 ) by pull down ( MI:0096 )

l MINT-7228466 : SRPK1 (uniprotkb: Q96SB4-2 ) physically interacts ( MI:0915 ) with SAFB1 (uniprotkb: Q15424 ) by anti tag coimmunoprecipitation ( MI:0007 )

l MINT-7228500 : SRPK1a (uniprotkb: Q96SB4-3 ) physically interacts ( MI:0915 ) with SAFB1 (uniprotkb: Q15424 ) by anti tag coimmunoprecipitation ( MI:0007 )

l MINT-7228483 : SRPK1 (uniprotkb: Q14151 ) physically interacts ( MI:0915 ) with SAFB2 (uni-protkb: Q14151 ) by anti tag coimmunoprecipitation ( MI:0007 )

Abbreviations

GFP, green fluorescent protein; GST, glutathione S-transferase; LBR, lamin B receptor; SAFB, scaffold attachment factor B; SRPK, SR protein kinase.

Although the SR protein kinase (SRPK) family was

discovered < 15 years ago, it has been implicated in

cellular processes of the utmost importance SRPKs

specifically phosphorylate serine residues in regions

rich in Ser⁄ Arg repeats, also called RS domains RS

domain-containing proteins are spread throughout the

cell They are functionally associated with a

multiplic-ity of cellular processes, such as splicing, pre-mRNA

processing, chromatin structure and remodeling,

tran-scription by RNA polymerase II, mRNA translation,

cell-cycle progression, cell structure and other

species-specific functions [1,2] The SRPK family of protein

kinases is highly conserved among eukaryotes, both

structurally and functionally [3–9]

The correct constitutive and alternative splicing, the

shuttling of several RS splicing factors between the

nucleus and the cytoplasm, their subnuclear

localiza-tion in nuclear speckles, their recruitment to sites of

transcription and their contribution to correct exon

selection via RNA binding, are some of the steps

regu-lated by SRPK phosphorylation [10–15] SRPKs have

also been implicated in mRNA export from and

pro-tein import into the nucleus [16,17] In addition,

phos-phorylation of the nucleoplasmic tail of the lamin B

receptor (LBR) by SRPK1 [18,19] regulates its binding

to chromatin [20] Protamine P1, a histone-replacing

protein is also phosphorylated by SRPK1 [21]

Phos-phorylation of the two proteins has been shown to

play a crucial role in mammalian spermiogenesis [22]

It is also interesting to note that several viruses alter

expression levels of SRPKs, and viral proteins interact

and become phosphorylated by SR kinases during the

infection cycle (human T-lymphotropic virus-1, herpes

simplex virus-1, hepatitis B virus), highlighting the

importance of the involvement of SRPKs in a large

number of cellular mechanisms [23–26]

In humans, the SRPK1 gene product is alternatively

spliced producing a minor transcript, the product of

which, SRPK1a, contains an additional 171 amino

acids at its N-terminus because of the retention of an

intron [27] In a previous study, which revealed the

expression of SRPK1a as an active kinase displaying

only minor differences from SRPK1, we showed that

its additional N-terminal region interacts with scaffold

attachment factor B1 (SAFB1) [27]

SAFB1 is a protein of the nuclear matrix first

dis-covered approximately a decade ago It was reported

with different names and was associated with a

diver-sity of functions [28–31] It is clear, however, that

SAFB1 resides in the nucleus and is a scaffold⁄ matrix

attachment region element binding protein It is 915

amino acids long and contains a SAF box (amino acids 35–67), a RNA recognition motif domain (amino acids 409–482), a putative nuclear localization signal (amino acids 519–614), a Glu⁄ Arg-rich region (amino acids 619–699) and a Gly-rich region (amino acids 785–899) A multiplicity of publications provide ample evidence that SAFB1 interacts with several different proteins such as polymerase II, splicing factors and hnRNP proteins and also with the tight junction pro-tein ZO-2 and the tripartite motif family propro-tein TRIM 27 [30,32–36] In addition, it is found in a num-ber of different subnuclear complexes formed by a variety of different combinations of nuclear proteins involved in either transcription [37] or splicing [38] The most prominent function of SAFB1, however, is transcriptional repression It was initially shown that SAFB1 binds to, and acts as, a corepressor of estrogen receptor a [39] It was further shown that it is capable

of suppressing the transcription of a reporter gene; suppression being exerted via interaction with TATA-binding protein associated factor II 68 [39] In addi-tion, SAFB1 represses the transcriptional activity of multiple nuclear receptors [40] This repression might

be effectuated, in some cases at least, by its interaction with the nuclear receptor corepressor which facilitates binding of histone deacetylases to the sites of tran-scription of nuclear receptors [41] SAFB2, a protein with 70% structural identity to SAFB1, is much less studied, but it also seems to be a transcriptional repressor [39,42] However, it cannot substitute for SAFB1, because SAFB1 knockout mice display serious defects [43] Moreover, SAFB1 and SAFB2 have dif-ferent subnuclear localizations [38] A third protein belonging to the SAFB family, SAF-like transcription modulator (34% identity to SAFB1 and 32% to SAFB2), has been shown to downregulate general mRNA synthesis [44] Finally, SAFB1, and the SAF-like transcription modulator, have been reported to exhibit pro-apoptotic activity [44,45]

Following our initial finding that the unique N-ter-minal part of SRPK1a interacts with SAFB1, we decided to investigate whether this interaction has an effect on the enzyme activity of the kinase In this study, we demonstrate that both kinases (SRPK1a and SRPK1) interact with both scaffold attachment factors (SAFB1 and SAFB2), albeit with different affinities Our in vitro experiments clearly show that this inter-action inhibits the activity of the kinases, whereas co-immunoprecipitation and subcellular fractionation analyses suggest that this inhibition also takes place

in vivo

FLAG–SRPK1a activity is inhibited in vitro by

SAFB1

SRPK1a interacts with SAFB1 via its N-terminus In

order to find out whether the interaction affects the

enzymatic activity of the kinase, we performed

phos-phorylation assays using as the substrate the

N-termi-nal 205 amino acids of LBR (NtLBR) in the

presence of increasing quantities of bacterially

expressed SAFB1 protein Because it was practically

impossible to obtain soluble recombinant SRPK1a

from bacteria, we used as a kinase source

immuno-precipitates of FLAG–SRPK1a from transfected

HeLa cell extracts [27] The SAFB1 protein used in

the assays was the bacterially produced glutathione

S-transferase (GST)-fused C-terminal amino acids 600–915 (GST–SAFB1C) [39] (Fig 1A) As shown in Fig 1B, SRPK1a phosphorylates bacterially produced GST–NtLBR (lane 1) and this phosphorylation is inhibited in a dose-responsive manner by the addition

of GST–SAFB1C (lanes 2–6) Moreover, the inhibi-tion is specific for SAFB1C because, when GST is added to the assay in quantities equal to those of GST–SAFB1C that completely inhibit the reaction, it does not affect the phosphorylation of GST–NtLBR (lane 7)

To verify that this inhibition is valid for more than one substrate, we used P2P-R, a nuclear matrix protein which contains RS motifs This protein has previously been shown to be phophorylated by SRPK1a [37] (and our unpublished observations) As shown in Fig 1C, SRPK1a phosphorylates bacterially

A

C

D

B

1 2

66-

45-

35-

25-FLAG–SRPK1a

FLAG–SRPK1a

GST–NtLBR

GST–P2P-R

GST–SAFB1C

GST–SAFB1C

FLAG–SRPK1a

R 0

FLAG–SRPK1a

GST–SAFB1C (µg) GST (µg)

R 0

GST–SAFB1C

FLAG–SRPK1a

+ +

+ +

+ +

+ +

– – – – 37.5

0 7.5 22.5 37.5 –

0 7.5 15 37.5 – – – – – 37.5

+ + + + + + +

+ + +

+ 0

– 7.5 – 15 – 22.5 – 30 – 37.5 – – 37.5

6 7

GST–NtLBR GST–SAFB1C (µg) GST (µg)

FLAG–SRPK1a GST–P2P-R GST–SAFB1C (µg) GST (µg)

1 2 3 4 5

1 2 3 4 5

1 2 3 4 5

6 7 1 2 3 4 5 6 GST

GST

GST

Fig 1 Effect of GST–SAFB1C on FLAG–SRPK1a activity (A) Bacterial preparation of GST–SAFB1C (lane 2) Full-length GST–SAFB1C is indi-cated by a dot Numbers indicate molecular mass in kDa (lane 1) (B) FLAG–SRPK1a kinase immunoprecipitated from HeLa whole-cell extract was incubated with GST–NtLBR and [ 32 P]ATP[cP] in the presence of GST–SAFB1C or GST (quantities were as indicated on the right), as described in Materials and methods Samples were analysed on 10% SDS ⁄ polyacrylamide gel Proteins were Coomassie stained (left) and labelled proteins were detected by autoradiography (right) (C) As in (B) except that GST–P2P-R(442–585) was used as a substrate (D) As in (B) except that peptide R0was used as a substrate and the SDS ⁄ polyacrylamide gel was 12%.

expressed GST–P2P-R(442–585), a fragment of the

P2P-R protein that contains a RS domain (lane 1)

and this phosphorylation is gradually abolished by

increasing amounts of GST–SAFB1C (Fig 1C,

lanes 2–4) Finally, when the artificial peptide R0

corresponding to the RS-rich LBR amino acid

sequence 70–91 was used as substrate [46], the results

were similar to the previous two experiments

(Fig 1D compare lane 1 with lanes 2–4), indicating

that SAFB1 inhibits phosphorylation by affecting the

kinase itself and not specific sequences on each of

the substrates

The RE domain of SAFB1 is not required for the

inhibition of FLAG–SRPK1a activity

Amino acids 619–699 of SAFB1 comprise its so-called

Glu⁄ Arg region, which is rich in RE dipeptides It has

been hypothesized that this region mimics

phosphory-lated RS dipeptides [47], the structure that the SRPK

substrates display after the phosphorylation reaction

In this context, we explored the possibility that this

region may play a role in the inhibition that SAFB1

exerts on SRPK1a activity To this end, we

con-structed a plasmid producing a fusion protein lacking

the RE-rich region, GST–SAFB1CDRE (amino acids

709–915) (Fig 2A) As shown in Fig 2B, the new

fusion protein still inhibits the phosphorylation of

GST–NtLBR by SRPK1a (Fig 2B, lanes 2–6) GST–

SAFB1CDRE also inhibits SRPK1a activity when

P2P-R or the R0 peptide is used as a substrate (data

not shown) These results show that the deleted

Glu⁄ Arg region of SAFB1 is not required for

inhibi-tion of SRPK1a activity

Along this line of thought, we tested SAFB2, the

close evolutionary relative of SAFB1 which also

con-tains the corresponding RE domain We constructed

the bacterially expressed fusion protein GST–SAFB2C

(Fig 2C) harboring amino acids 641–953 of the

C-ter-minal region of SAFB2, which corresponds to the

respective sequences of GST–SAFB1C, including the

Glu⁄ Arg region

As shown in Fig 2D, GST–SAFB2C is practically

unable to inhibit the phosphorylation of GST–NtLBR

by SRPK1a (compare lane 1 with lanes 2–5) and the

barely detectable inhibition is in quantities of GST–

SAFB2C significantly exceeding those of GST–

SAFB1C (or GST–SAFB1CDRE) that totally inhibit

SRPK1a activity (lanes 5–6) These results were

con-firmed using P2P-R and the R0 peptide as substrates,

as shown in Fig 2E,F These data confirm the inability

of the RE-rich region to inhibit the phosphorylating

activity of SRPK1a

FLAG–SRPK1 activity is inhibited in vitro by SAFB1 and SAFB2

We next asked whether the inhibition exerted by SAFB1 on SRPK1a activity is because of its inter-action with the N-terminal part of the kinase We approached this question indirectly by examining the effect of GST–SAFB1C, GST–SAFB1CDRE and GST–SAFB2C on SRPK1, which in its full-length is 100% identical to the SRPK1a molecule, except for the absence of amino acids 5–174 SRPK1 was expressed, like SRPK1a, as a FLAG-tagged protein

in HeLa cells, immunoprecipitated by the M2 mono-clonal anti-FLAG IgG and used as such, in in vitro phosphorylation assays, with GST–NtLBR as the substrate

As shown in Fig 3A (lane 1, compare with lanes 2–6), SRPK1 activity is inhibited by GST–SAFB1C to

a similar extent to the inhibition exerted on SRPK1a Accordingly, GST–SAFB1CDRE also inactivates FLAG–SRPK1 to the same extent as FLAG–SRPK1a (Fig 3B) However, as shown in Fig 3C (compare lane 1 with lanes 2–6), FLAG–SRPK1 activity is also clearly inhibited by GST–SAFB2C, unlike that of FLAG–SRPK1a (Fig 2D)

Bacterially expressed GST–SRPK1 activity is inhibited in vitro by SAFB1 and SAFB2 The fact that SAFB1C inhibits both SRPK1a and SRPK1 suggests that it does not bind to SRPK1a only via its unique N-terminal part, but also via other regions common to the two kinases However, because

in our assays we always used kinases immunoprecipi-tated from whole-cell extracts we decided to rule out the possibility that a third protein intervenes in the SRPK1⁄ 1a–SAFB1 ⁄ 2 interaction To this end, we pre-pared bacterially expressed GST–SRPK1 which is rela-tively easily purified [19], (Fig 4A) and used it in

in vitro phosphorylation assays (0.7 lg of GST fusion protein per assay) with GST–Nt-LBR as the substrate and increasing quantities of each of the three fusion proteins GST–SAFB1C, GST–SAFB1CDRE and GST–SAFB2C As shown in Fig 4B, the recombinant kinase was active in phosphorylating GST–NtLBR (lane 1) The results are practically identical to those obtained with the whole-cell extract

immunoprecipitat-ed kinase, because when GST–SAFB1C is includimmunoprecipitat-ed in the phosphorylation assay at increasing quantities, phosphorylation is gradually abolished (lanes 2–6)

As expected, GST–SRPK1 is inhibited by GST–SAFB1 DRE (Fig 4C) and also by GST–SAFB2C (Fig 4D),

as observed in the case of HeLa cell isolated kinase Up

to 500 ng of full-length GST–SRPK1 tested in our

experiments was inhibited by the maximal quantity of

GST–SAFB1C used in all the assays (data not shown)

These results indicate that the inhibition exerted by

SAFB1 and SAFB2 on SRPK1 is caused by a direct

interaction between each of the two proteins and

SRPK1 In an attempt to map the region of this

inter-action on the kinase, we also produced in bacteria a

truncated form of GST–SRPK1, from which amino

acids 256 to 475, containing the so-called spacer region

of the kinase, are deleted (GST–SRPK1Dspacer) It has previously been shown that removal of the spacer domain has no effect on catalytic activity but drasti-cally affects the subcellular localization of the kinase [48] Indeed, as shown in Fig 4E, GST–SRPK1Dspacer

is able to phosphorylate its substrate, GST–NtLBR, as efficiently as GST–SRPK1 (lanes 1 and 5) In addition, its activity is inhibited by both SAFB1 and SAFB2 (lanes 6 and 7), implying binding of the SAFB proteins

to (a) region(s) other than the spacer region

C

E

F

D

1

+ + + + + + + + + + +

+ + + + + + + +

+ +

+ + + +

+ +

+ + + +

0 7.5 22.5 37.5

+ + + +

+ + + +

0 7.5 22.5 37.5

GST–NtLBR

GST–NtLBR

GST–SAFB1C ΔRE

FLAG–SRPK1a

GST–NtLBR

GST–SAFB2C

FLAG–SRPK1a

GST–P2P-R

GST–SAFB2C

FLAG–SRPK1a GST–P2P-R GST–SAFB2C (µg)

FLAG–SRPK1a

R 0 GST–SAFB2C (µg)

FLAG–SRPK1a

R 0

GST–SAFB2C

1 2 3 4

1 2 3 4

FLAG–SRPK1a GST–NtLBR GST–SAFB2C (µg)

66-

45-

35-

25-

116-

66-

45-

35-

25-1 2 2

Fig 2 Effect of GST–SAFB1CDRE and GST–SAFB2C on FLAG–SRPK1a activity (A) Bacterial preparation of GST–SAFB1CDRE (lane 2) Full-length GST–SAFB1CDRE is indicated by a dot Numbers indicate molecular mass markers in kDa (lane 1) (B) FLAG–SRPK1a kinase immuno-precipitated from HeLa whole-cell extracts was incubated with GST–NtLBR and [ 32 P]ATP[cP] in the presence of GST–SAFB1CDRE (quantities were as indicated in the right panel) as described in Materials and methods Samples were analysed on 10% SDS ⁄ polyacrylamide gels Pro-teins were Coomassie stained (left) and labelled proPro-teins were detected by autoradiography (right) (C) Bacterial preparation of GST–SAFB2C (lane 2) Full-length GST–SAFB2C is indicated by a dot Numbers indicate molecular mass markers in kDa (lane 1) (D) As in (B) except that the indicated quantities of GST–SAFB2C were used (E) As in (D) except that GST–P2P-R(442–585) was used as a substrate (F) As in (D) except that peptide R 0 was used as a substrate and the SDS ⁄ polyacrylamide gel was 12%.

Both SRPK1, as well as SRPK1a, bind to both

SAFB1 and SAFB2

In order to confirm SAFB1 and SAFB2 binding to the

kinases, we performed an affinity chromatography

experiment in which we immobilized M2

antibody-bound FLAG–SRPK1 or SRPK1a on beads and

incu-bated them with GST–SAFB1C, GST–SAFB1CDRE,

GSTSAFB2C and GST bacterial preparations The

beads were washed and the eluted proteins were

probed with the anti-GST IgG for SAFB protein

detection and the anti-FLAG IgG for the

immuno-precipitated SRPK1 protein calibration

As shown inFig 5, both SAFB proteins bind clearly

and specifically to both kinases (no proteins bind to

beads alone: lanes 9, 10, 11 and 12), albeit with

differ-ent affinities SAFB1 binds tightly to SRPK1a and

almost as tightly to SRPK1 (compare lanes 2 and 1)

and the same holds true for the GST–SAFB1CDRE

protein (lanes 3 and 4), whereas SAFB2 barely binds

to SRPK1a (lane 6) but almost as tightly as SAFB1 to

SRPK1 (compare lanes 5 and 1) These results confirm

the direct interaction of SAFBs with the SRPK1⁄ 1a

proteins and provide a direct link between the affinity

of the SAFB–SRPK1⁄ 1a interaction and the extent of the inhibition exerted on kinase activity in each case (see Discussion) Also, both GST–SAFB1 and GST– SAFB2 were able to bind to a FLAG–SRPK1DSpacer fusion protein in a similar experiment, suggesting that the catalytic region of the kinase, comprising amino acids 1–256 and 476–655, interacts with the SAFB proteins (data not shown)

In SRPK⁄ SAFB complexes, able to form in living cells, SRPK activity is inhibited

In a following step, we asked whether the corresponding SRPK⁄ SAFB complexes were able to form in living cells with full-length SAFB proteins, and whether the kinases

in these complexes were also repressed HeLa cells were co-transfected with plasmids expressing FLAG, FLAG– SRPK1a or FLAG–SRPK1 together with green fluores-cent protein (GFP), GFP–SAFB1 or GFP–SAFB2, lysed and FLAG proteins were immunoprecipitated with the M2 anti-FLAG IgG Kinase assays were performed on the immunoprecipitated kinases using

A

B

C

FLAG–SRPK1

GST–NtLBR

GST–SAFB1C

FLAG–SRPK1

GST–NtLBR

GST–SAFB1CΔRE

GST–SAFB1CΔRE (µg)

0 7.5 15 22.5 30 37.5

0 7.5 15 22.5 30 37.5

0 7.5 15 22.5 30 37.5

GST–NtLBR GST–SAFB1C (µg)

FLAG–SRPK1 GST–NtLBR

GST–SAFB2C (µg)

FLAG–SRPK1 GST–NtLBR

FLAG–SRPK1

GST–NtLBR

GST–SAFB2C

Fig 3 Effect of GST–SAFB1C, GST–SAFB1CDRE and GST–SAFB2C on FLAG–SRPK1 activity (A) FLAG–SRPK1 kinase immunoprecipitated from HeLa whole-cell extracts was incubated with GST–NtLBR and [ 32 P]ATP[cP] in the presence of GST–SAFB1C (quantities were as indi-cated in the right panel) as described in Materials and methods Samples were analysed on 10% SDS ⁄ polyacrylamide gels Proteins were Coomassie stained (left) and labelled proteins were detected by autoradiography (right) (B) As in (A) except that the indicated quantities of GST–SAFB1CDRE were used (C) As in (A) except that the indicated quantities of GST–SAFB2C were used.

GST–NtLBR as a substrate In order to monitor the

SRPK⁄ SAFB complex assembly, half of the extract was

used to detect the proteins with the anti-FLAG and the

anti-GFP IgG As shown in Fig 6A both kinases are

active when extracted from cells co-expressing GFP

(lanes 1 and 4) However, when SRPK1 is co-expressed

with either SAFB1 or SAFB2, its activity is clearly

inhibited (lanes 2, 3) SRPK1a activity is also inhibited

by SAFB1 (lane 5), yet inhibition by SAFB2 is much weaker (lane 6)

As shown in Fig 6B, both GFP–SAFB1 and GFP–SAFB2 proteins bind to both kinases (left and right panels, lanes 2, 3 and 5, 6), unlike GFP (lanes

1 and 4) which does not bind by itself However,

A

C

D

E

B

1 2

116- GST–SRPK1

GST–NtLBR

GST–SAFB1C

GST–SRPK1

GST–NtLBR

GST–SAFB1C ΔRE

GST–SRPK1

GST–NtLBR

GST–SAFB2C

GST–SAFB1C ΔRE (µg)

GST–SRPK1 GST–NtLBR GST–SAFB1C (µg)

GST–SRPK1 GST–NtLBR

GST–SRPK1 Δspacer

GST–SAFB1C (µg) GST–SAFB2C (µg) GST

GST–NtLBR

GST–SRPK1 + + + +

+ + + + + +

37.5 37.5 37.5

37.5

37.5 37.5

+ + + + + + – –

– – –

– – – – –

– –

– – –

– – – –

– –

GST–SAFB2C (µg)

GST–SRPK1 GST–NtLBR

GST

+ + + + + + + + + + + + + +

+ + + + + + + + + + + +

– – – – – – 37.5

7

6 5 4 3 2 1

6 7 8 5

4 3 2 1

6 5 4 3 2 1

6 5 4 3 2 1

0 7.5 15 22.5 30 37.5 –

0 7.5 15 22.5 30 37.5

+ + + + + + + + + + + +

0 7.5 15 22.5 30 37.5

1 2 3 4 5 6 7

1 2 3 4 5 6

1 2 3 4 5 6

GST

66-

45-

35-

25-Fig 4 Effect of GST–SAFB1C, GST–SAFB1CDRE and GST–SAFB2C on GST–SRPK1 and GST–SRPK1Dspacer activity (A) Bacterial prepara-tion of GST–SRPK1 (lane 2) Full-length GST–SRPK1 is indicated with a dot Numbers indicate molecular mass markers in kDa (lane 1) (B) GST–SRPK1 purified from Escherichia coli whole-cell extract was incubated with GST–NtLBR and [32P]ATP[cP] in the presence of GST– SAFB1C or GST (quantities were as indicated in the right panel) as described in Materials and methods Samples were analysed on 10% SDS ⁄ polyacrylamide gels Proteins were Coomassie stained (left) and labelled proteins were detected by autoradiography (right) (C) As in (B) except that the indicated quantities of GST–SAFB1CDRE were used (D) As in (B) except that the indicated quantities of GST–SAFB2C were used (E) GST–SRPK1 (lanes 1–4) and GST–SRPK1Dspacer (lanes 5-8) purified from E coli whole-cell extract were incubated with GST–NtLBR and [ 32 P]ATP[cP] in the presence of the indicated quantities of GST–SAFB1C, GST–SAFB1C or GST as described in (B) Labelled proteins were detected by autoradiography.

although SAFB1 seems to bind to SRPK1 almost as

well as to SRPK1a, SAFB2, which is clearly present

in the eluate of SRPK1, is barely detectable in the

eluate of SRPK1a These results show that SRPK⁄

SAFB complexes are able to form in living cells and

that cell-extracted SAFB-bound kinases are inactive

in vitro

SRPK⁄ SAFB complexes are present in the nucleus

of HeLa cells Finally, we sought to detect the existence of endogenous SRPK⁄ SAFB complexes SAFB proteins were immuno-precipitated from whole-cell lysates of exponentially growing HeLa cells and the eluate was probed with the

Input

GST–

SAFB1C

GST–

SAFB1C

GST–

SAFB1C ΔRE

GST–

SAFB1C ΔRE

GST–

SAFB2C

GST–

SAFB2C

GST

GST GST–

SAFB1C

GST–

Anti-GST

Anti-FLAG

Anti-GST

1 2 3 4 5 6 7 8 9 10 11 12

Eluate

Beads

Fig 5 Binding of the GST–SAFB1/2

proteins on immobilized FLAG–SRPK1/1a

proteins FLAG–SRPK1 and FLAG–SRPK1a

were immunoprecipitated on beads by from

HeLa cell extracts and the beads were

incubated with GST–SAFB1C, GST–

SAFB1CDRE, GST–SAFB2C or GST The

eluates were subjected to electrophoresis

and proteins were detected using the

anti-FLAG and the anti-GST IgG as indicated

(lanes 1-8) The GST fusion proteins were

also incubated with beads alone treated

with immunoprecipitates from whole-cell

extracts of non-transfected cells (lanes

9–12) (lower) One-thirtieth of the input

proteins were subjected to electrophoresis

and detected using the anti-GST IgG

(upper).

A

B

GST–NtLBR FLAG–SRPK1 FLAG–SRPK1a GFP

GFP–SAFB1 GFP–SAFB2

Anti-GFP

Anti-FLAG

Anti-GFP

SAFB1

GFP–

SAFB2

SAFB1

GFP– SAFB2

SAFB1

GFP–

+ + – + – –

+ + – – + –

+ + – – – +

+ – + + – –

+ – + – + –

+ – + – – +

Fig 6 The binding effect of full-length

SAFB1 and SAFB2 on SRPK1 and SRPK1a

kinase activities HeLa cells were

co-trans-fected with plasmids expressing FLAG,

FLAG–SRPK1 or FLAG–SRPK1a together

with GFP, GFP–SAFB1 or GFP–SAFB2.

Whole-cell extracts were

immunoprecipitat-ed with an anti-FLAG IgG A) On half of the

immunoprecipitated material a kinase assay

was performed, as described in Materials

and methods Samples from the kinase

assay were analysed on SDS ⁄

polyacryl-amide gel and labelled proteins were

detected by autoradiography (B) The

remaining half of the immunoprecipitates

were analysed on SDS ⁄ polyacrylamide gel

and immunoblotted with an anti-FLAG and

an anti-GFP IgG On the same gel, 1 ⁄ 10 of

the quantity of the whole-cell extract that

was used in each immunoprecipitation

assay was subjected to electrophoresis and

immunoblotted with an anti-GFP IgG.

anti-SRPK1 IgG (the mAbs available do not distinguish

between SAFB1⁄ 2 and SRPK1 ⁄ 1a) An

immunoreac-tive band was detected in the eluate of SAFB

co-immu-noprecipitated proteins (Fig 7, lane 4), indicating that

part of endogenous SRPK1⁄ 1a is complexed with

SAFB in HeLa cells Approximately 2% of the total

SRPK1⁄ 1a levels was calculated to

co-immunopercipi-tate with the SAFB proteins (based on the intensinty of

the bands on the western blot) In order to determine in

which subcellular compartment such complexes may

form, we proceeded in subcellular fractionation of

HeLa cells and subsequent immunoblotting of the

frac-tions with the anti-SAFB and anti-SRPK1 IgG (Fig 8)

As expected, SAFB is detected in the nucleus where

it is found mostly in the nuclear matrix ( 80%),

whereas SRPK1⁄ 1a is detected mainly in the cytoplasm

( 60%) and to a lesser extent in the nucleus ( 40%)

Notably, a small but clearly detectable fraction of the

kinase ( 10%) is found in the nuclear matrix where

the SAFB concentration is high (Fig 8A)

Conse-quently, SRPK⁄ SAFB-containing complexes may exist

in the nuclear matrix as well as in the nucleoplasm

When the different fractions were assayed for SR kinase

activity using GST–NtLBR as a substrate,

phosphory-lation was easily detected in the cytoplasmic and

nucle-oplasmic fractions, but none was detected in the nuclear

matrix (Fig 8B), despite the presence of SRPK1⁄ 1a

molecules in this fraction (Fig 8A)

Discussion

In this study, we have followed up our initial

observa-tion that SRPK1a, the alternatively spliced form of

SRPK1, interacts with the nuclear matrix protein SAFB1 via its unique additional N-teminal domain

In our pursuit of a biological consequence of this interaction, we examined the activity of SRPK1a in the presence of SAFB1 and showed that the kinase is inhibited by this factor in vitro In our assays, we used the C-terminal region of SAFB1 because it includes the area found to interact with SRPK1a (amino acids 585–720) [27] Inhibition was evident when SRPK1a activity was tested on three different substrates (LBR, P2P-R and a RS domain-containing synthetic peptide) eliminating the possibility that SAFB1 interferes with different domains in each substrate However, rela-tively large quantities ( 20 lg) of our total bacterial GST–SAFB1C preparation were needed to eliminate phosphorylation of the substrates We cannot be

WB: a-SRPK1

IP: a-SAFB 127

77

Input (20%)

Fig 7 A fraction of endogenous SRPK1 ⁄ 1a co-immunoprecipitates

with SAFB1 ⁄ 2 Complexes between SAFB and SRPK1 ⁄ 1a proteins

were immunoprecipitated from HeLa cell extracts with a monoclonal

anti-HET ⁄ SAFB IgG and analysed on 10% SDS ⁄ polyacrylamide gels.

The proteins were then transferred to nitrocellulose and SRPK1 ⁄ 1a

was detected with the monoclonal anti-SRPK1 IgG, recognizing both

isoforms (lane 4) No direct immunoprecipitation of SRPKs was

observed when an irrelevant monoclonal anti-GFP IgG was used as

control (lane 3) A standard amount of cell extract, one-fifth of which

is shown (lane 2), was used in each immunoprecipitation assay.

Molecular mass markers are shown in kDa on the left.

A

B

Fig 8 Distribution of endogenous SAFB1 ⁄ 2 and SRPK1 ⁄ 1a pro-teins in HeLa cells, following biochemical fractionation (A) The dis-tribution of SAFB1 ⁄ 2 and SRPK1 ⁄ 1a proteins between the various fractions was analysed by immunoblotting using a mouse mono-clonal anti-HET/SAFB and a monomono-clonal anti-SRPK1 IgG respectively (the available antibodies do not distinguish between SAFB1 ⁄ 2 and SRPK1 ⁄ 1a, respectively; see Materials and methods for the analyti-cal fractionation protocol) (B) The different fractions were assayed for RS kinase activity, using bacterially produced GST–NtLBR as substrate The samples were analysed by SDS ⁄ PAGE and auto-radiographed The radioactive bands corresponding to labelled GST–NtLBR from were excised, and the radioactivity was deter-mined by Cerenkov counting RS kinase activity of the different fractions is expressed as total units (%).

certain whether this is because only the full-length

SAFB1C, which is a relatively small fraction in the

total population of the GST-purified peptide, inhibits

the kinase or because SAFB1 homopolymerizes via its

RE-rich region [38,42] (and our unpublished

observa-tions), large quantities are needed to have sufficient

SAFB1 monomers available to form complexes with

the kinase under the conditions of the in vitro

phos-phorylation assays

The Glu⁄ Arg-rich region is a candidate for

repres-sion of the kinase activity because of its particular

structure, which resembles a phosphorylated RS

domain RS domains are known to be protein–protein

and protein–RNA interaction surfaces and the

phos-phorylation of serines affects those interactions as well

as interactions between RS domain-containing proteins

[49,50] Very recently the core complex of SRPK1 with

one of its substrates, the spliceosome factor ASF⁄ SF2,

has been crystallized [51], revealing important

informa-tion and confirming previous observainforma-tions [52–54]

about the significance of enzyme–substrate contacts for

catalysis This study highlights the importance of the

interaction of a phosphoserine of the RS domain of

the substrate with the catalytic region of the kinase, on

the one hand, and of the positive charge of the RS

domain with the negatively charged ‘docking groove’

of SRPK1, on the other hand The RE-rich region of

SAFB1 may interfere with any of these processes by

mimicking a phosphorylated RS domain, thus

disturb-ing the catalytic activity of the kinase However, when

we deleted the sequence rich in RE dipeptides, the

remaining SAFB1 sequences (709–915) still inhibited

SRPK1a activity, implying that some particular

struc-tural element or configuration in this region should be

responsible for the observed effect At this point, it

should be noted that among the SAFB1 sequences

contained in the GST–SAFB1CDRE fusion protein,

several RE dipeptides are scattered so that we cannot

exclude the possibility that inhibition is exerted by

these sequences, particularly because the pI of the

remaining SAFB1 sequence in the GST–SAFB1CDRE

peptide is still basic Otherwise, it could be a

combina-tion of two effects, where both specific structural

ele-ments and the scattered REs would contribute to the

observed inhibition

We were intrigued to find that, although SAFB1

interacts with the unique N-terminal part of SRPK1a,

it also inhibits the activity of SRPK1 which lacks this

unique part We excluded the possible involvement of

a cellular protein in our in vitro assays by using

bacte-rially purified GST-fused SRPK1 and confirmed that

both SAFB1 and SAFB2 repress its activity Affinity

chromatography experiments confirmed that SAFB1

and SAFB2 bind to SRPK1, which suggests that there exists a domain on the kinase recognized by each of the two factors We excluded the possibility that such

a domain is located in the spacer region of the kinase because both SAFB proteins still bind on a kinase molecule from which the spacer domain is deleted and they still inhibit its enzymatic activity Thus, the inhi-bition mechanism involves binding of the SAFB mole-cules to the catalytic domain of SRPK1 Furthermore, SAFB1 also binds to SRPK1a (as previously shown), but SAFB2 barely does We obtained the same quali-tative results concerning the relative affinities of the four proteins when we co-immunoprecipitated the kin-ases with SAFB proteins from HeLa cells This result

is not very easy to explain for the pair SRPK1a– SAFB2 Because SAFB2 interacts with SRPK1, it must recognize and bind to a specific region on it, which is evidently also contained in SRPK1a How-ever, SAFB2 binds only weakly to SRPK1a One should then accept that either SAFB1 and SAFB2 have two different, though almost equal in strength, types of interaction with SRPK1 that are differenti-ated in SRPK1a because of the N-terminal domain,

or that even if their interaction with SRPK1 is simi-lar, the N-terminal domain has a stabilizing effect on SAFB1, but a destabilizing effect on the interaction with SAFB2 In any case, additional dissection of the SAFB and SRPK1 molecules is required to determine the regions responsible for their interactions It should

be pointed out, however, that this study is the first to reveal functional differences between SAFB1 and SAFB2

Although at this stage in our study our results can only be qualitative, we have noticed that the inhibit-ing activity exerted by SAFB proteins on the kinases,

is closely related to the affinity with which they inter-act in affinity chromatography and co-immunoprecipi-tation experiments Although this is not unexpected, it

is indicative of the importance of SRPK⁄ SAFB intra-cellular complexing We were able to demonstrate the existence of SRPK⁄ SAFB complexes in HeLa cells, estimating the percentage of total cellular SRPK1⁄ 1a molecules occupied in these complexes to be  2% Because the antibodies used cannot distinguish between SRPK1 and SRPK1a, or between SAFB1 and SAFB2, we do not know the exact composition

of the detected SRPK⁄ SAFB complexes However, using overexpressed proteins in HeLa cells we demon-strated that all four complexes, SRPK1a–SAFB1, SRPK1–SAFB1, SRPK1–SAFB2 and SRPK1a– SAFB2 (listed by relative order of affinity) can be formed in living cells by the full-length proteins Moreover, SRPK1⁄ 1a molecules extracted from the