The RING Domain of RAG1 Ubiquitylates Histone H3 - Cell Press

Molecular Cell

Article The RING Domain of RAG1 Ubiquitylates Histone H3: A Novel Activity in Chromatin-Mediated Regulation of V(D)J Joining Ursula Grazini,1,3,4 Federica Zanardi,2 Elisabetta Citterio,2,5 Stefano Casola,2 Colin R. Goding,3,4 and Fraser McBlane1,* 1Department

of Experimental Oncology, European Institute of Oncology, Milan 20141, Italy FIRC di Oncologia Molecolare, 20139 Milan, Italy 3Marie Curie Research Institute, Oxted, Surrey RH8 OTL, UK 4Ludwig Institute for Cancer Research, University of Oxford, Oxford OX3 7DQ, UK 5Present address: Division of Molecular Genetics, The Netherlands Cancer Institute, 1066 CX Amsterdam, The Netherlands *Correspondence: [email protected] DOI 10.1016/j.molcel.2009.12.035 2Istituto

SUMMARY

The RAG1 and RAG2 proteins are the only lymphoidspecific factors required to perform the first step of V(D)J recombination, DNA cleavage. While the catalytic domain of RAG1, the core region, has been well characterized, the role of the noncore region in modulating chromosomal V(D)J recombination efficiency remains ill defined. Recent studies have highlighted the role of chromatin structure in regulation of V(D)J recombination. Here we show that RAG1 itself, through a RING domain within its N-terminal noncore region, preferentially interacts directly with and promotes monoubiquitylation of histone H3. Mutations affecting the RAG1 RING domain reduce histone H3 monoubiquitylation activity, decrease V(D)J recombination activity in vivo, reduce formation of both signal-joint and coding-joint products on episomal substrates, and decrease efficiency of V(D)J recombination at the endogenous IgH locus in lymphoid cells. The results reveal that RAG1-mediated histone monoubiquitylation activity plays a role in regulating the joining phase of chromosomal V(D)J recombination.

INTRODUCTION The antigen receptor repertoire is generated during lymphocyte development by V(D)J recombination, a complex chromosomal reaction that assembles antigen-specific immunoglobulin (Ig) and T cell receptor (TCR) genes into a functional coding unit by the imprecise joining of multiple V, D, and J gene segments. The reaction is targeted by recombination signal sequences (RSSs) that flank all rearranging segments. The products of the recombination activating genes 1 and 2 (RAG1 and RAG2) initiate recombination by generating double-strand DNA breaks between the RSS and the flanking gene segments (van Gent

et al., 1995). This generates a pair of RSS signal ends and a pair of coding ends. After end processing, the nonhomologous end-joining (NHEJ) repair machinery joins the reassembled ends into coding joints and signal joints. Both RAG1 and RAG2 are essential for V(D)J recombination and hence the production of B and T lymphocytes (Mombaerts et al., 1992; Shinkai et al., 1992). Additionally, both RAG proteins are essential and sufficient to initiate V(D)J recombination in vitro (McBlane et al., 1995). The RAG proteins assemble 12-RSS and 23-RSS into a synaptic complex, in which complete double-strand cleavage occurs at each signal. After cleavage, RAG proteins remain stably bound to the signal ends in a postcleavage complex (Agrawal and Schatz, 1997) before the joining reaction ultimately occurs. In addition to the functionally essential ‘‘core regions’’ of RAG1 (aa 384–1008) and RAG2 (aa 1–387), several studies suggest a potential role of the noncore RAG regions in nonenzymatic aspects of the reaction, including end processing and joining, and regulation of RAG protein stability (Talukder et al., 2004). A role for the N-terminal region of RAG1 in modulating V(D)J recombination through a chromatin-mediated mechanism was suggested by the altered recombination efficiency of RAG1 mutants at specific loci (Dudley et al., 2003; Noordzij et al., 2000; Santagata et al., 2000). The influence of chromatin structure on V(D)J recombination activity is well established, although the molecular mechanisms remain obscure. Most studies have focused on chromatin regulation of the RAG-mediated RSS cutting step (Golding et al., 1999; Kwon et al., 1998; McBlane and Boyes, 2000), although how the broken ends are repaired within a chromatin context remains unknown. The N-terminal region of RAG1 contains a conserved RING finger that defines a growing subset of ubiquitin ligases (Jackson et al., 2000). Modification of proteins by ubiquitin conjugation (ubiquitylation) influences many aspects of cell biology, including proteosomal degradation, signaling, cell-cycle control, differentiation, and DNA repair (Weissman, 2001). The RING finger domain within the N-terminal region of RAG1 (aa 264–389) acts as an E3 ubiquitin ligase in vitro (Yurchenko et al., 2003) and mediates auto-ubiquitylation in cells (Jones and Gellert, 2003). Although a putative substrate, the nuclear transport protein

282 Molecular Cell 37, 282–293, January 29, 2010 ª2010 Elsevier Inc.

Molecular Cell RAG1 Ubiquitylates Histone H3

Figure 1. N-Terminal RAG1 Interacts with Endogenous Histone H3 in 293 Cells

A

11

3383 83

11008 008 11040 040

N-term

RAG1 core

G

1

1

A

B

G

G

LA

(1

-1

0 04

F

LA

F

G

LA

F

RAG1 FL

WB: α : FLAG

WB: α : FLAG

IgH H3

WB α : H3

WB α : H3

H3 tot 1

2

) 83

R

3

G

C

F

)R

A

)R

LA

G

A

1

G

1

2

karyonpherin alpha 1 (KPNA1), for the RAG1 ubiquitin ligase activity has been identified in vitro (Simkus et al., 2009), additional targets for the ubiquitylation activity of RAG1 in vivo are unknown. Here we demonstrate that RAG1, through its N-terminal RING domain, interacts with and promotes monoubiquitylation of histone H3. Importantly, mutations disrupting RAG-mediated H3 ubiquitylation reduce the efficiency of the joining step of V(D)J recombination in vivo, suggesting that H3 ubiquitylation by RAG1 may represent an important link between the cutting and joining steps of this chromosomal recombination reaction.

RESULTS N-Terminal RAG1 Binds Histones In Vitro and In Vivo Biochemical studies on RAG proteins have identified catalytically active regions of each protein, referred to as the core regions, which include residues 384–1008 for RAG1 (Figure 1A) and 1–387 for RAG2. Core RAG1, together with core RAG2, retains all DNA cleavage activity in vivo and in vitro (Gellert, 2002). However, noncore regions of RAG1 and RAG2, although not required for catalysis of the initial cleavage step, are evolutionarily conserved and can influence reaction efficiency by

(1

G

08

10

(A) Schematic representation of full-length murine RAG1 protein. N-terminal region (aa 1–383) and core region (aa 384–1008) of RAG1 are represented as bars. (B and C) Constructs encoding the indicated RAG1 sequences were transfected in 293 cells and proteins immunoprecipitated with anti-FLAG antibody, washed under stringent conditions, separated by 10% SDS-PAGE (B) or 15% SDSPAGE (C) and immunoblotted with anti-FLAG antibody or with anti-histone H3 antibody, as indicated on the left of each panel. A fraction (2.5%) of input material is also shown (H3 tot.).

84

(3

LA

F

serving important regulatory functions in chromosomal V(D)J recombination (OetRAG1 core tinger, 2004). Several lines of evidence also indicate a potential role of the N-terminal region of RAG1 in modulating IgH V(D)J recombination through a chromatin-mediated mechanism (Noordzij RAG1 N-term. et al., 2000; Santagata et al., 2000). Therefore, we investigated whether the N-terminal RAG1 region contains a histone binding region. We first tested whether endogenous histones coimmuH3 noprecipitate with ectopically expressed RAG1 protein. Constructs encoding H3 tot FLAG-(1–1040)RAG1 full-length, or FLAG3 (1–383)RAG1 N-terminal, or FLAG-(3841008)RAG1 core were transfected into 293 cells, proteins immunoprecipitated 48 hr later with anti-FLAG antibody, and histone binding assayed by western blotting with anti-histone H3 antibody as probe. Endogenous histone H3 was coimmunoprecipitated with RAG1FL (Figure 1B, lane 2). As shown in Figure 1C, histone interaction was observed with the N-terminal region of RAG1 (lane 2), but not with the core region (lane 3), indicating that RAG1 protein interacts with H3 through its N-terminal region. A construct encoding FLAG alone was included as negative control (lanes 1). RAG1 still coprecipitated with histone H3 after DNase treatment of transfected nuclei, thus excluding a DNA-mediated interaction (see Figure S1 available online). We next performed an in vitro GST pull-down assay, to assess whether the interaction between N-terminal RAG1 and histones was direct. GST-(1– 1040)RAG1FL, GST-(1–383)RAG1 N-terminal, and GST-(383– 1008)RAG1 core proteins were prepared, incubated with purified histones assembled into octamers (Figure 2A), and precipitated with glutathione Sepharose 4B beads. Histone coprecipitation was detected by western blot analysis using an anti-H3 antibody. Histone H3 was coprecipitated with GST-RAG1FL (Figure 2A, lane 2) and the RAG1 N-terminal region (lane 4), but not with the core region (lane 3) or GST alone (lane 1). This result confirms that RAG1 interacts directly with H3 through its N-terminal region. Using recombinant histones, a subsequent far-western analysis revealed that N-terminal RAG1 displays

Molecular Cell 37, 282–293, January 29, 2010 ª2010 Elsevier Inc. 283


ST -R A G ST G1

FL

B

+

+

+

1

2

M 1

2

G G ST ST GGS STT -R GG -RAAG SST G 1 T1 G (-3(388 FFLL GST 44ST-( 1-10 -1(1-3 0008 -383 8)R)R 83)R AAG )RA G11 AGG1 1

+

G

G

ST -R A G ST G1

A

FL

Figure 2. RAG1 Interacts with Histones In Vitro through Its N-Terminal Region

2

3

4

M

WB: a-H3 -H3

Ponceau staining

FL

C

WB: -GST

FL

1

(A) Equal amounts (5 mg each) of the indicated GST proteins were incubated with 3.5 mg of purified histones assembled into octamers (A) for 2 hr at 4 C in binding buffer and precipitated with glutathione Sepharose 4B beads. After extensive washing, proteins were separated by 15% SDSPAGE. Histone H3 coprecipitation was detected by western blotting using an anti-H3 antibody. Input in vitro-translated (IVT) histones (10%) are shown (B, left panel). (B and C) Far-western analysis of recombinant full-length histones (B) or recombinant full-length (FL) and tailless histone H3 proteins (C), 2.5 mg of histones were separated by 16% SDS-PAGE, transferred to nitrocellulose membrane, and Ponceau-stained (B and C, left panels). After overnight incubation at 4 C with 2.5 mg of GST or GST-RAG1 full-length (as indicated at the top) and extensive washing, proteins bound to histone-associated membrane were analyzed by probing with anti-GST antibody (right panels). Marker sizes (M) are expressed in kilodaltons.

1

AG

ST -R

G

+

ST

G

+

ST -R

G

+

+

G

ST

AG

1

(lanes 2 and 4) histone H3 were separated by 16% SDS-PAGE electrophoresis, transferred to nitrocellulose membrane, Ponceau stained (left panel), and probed using purified GST or GSTRAG1 (as indicated at the top). After extensive washing, protein bound to membrane-associated histone H3 was (FL) H3 analyzed by western blotting analysis (tailless) H3 using an anti-histone H3 antibody (right panel). Tailless histones could not bind to purified GST-RAG1 (lane 4), while M 1 2 M 3 4 M 1 2 M 3 4 full-length control histone H3 bound efficiently (lane 3). No signals were detected Ponceau staining WB: -GST on membrane-associated full-length or tailless histone H3 incubated with GST a strongly preferential association with histone H3. As shown in alone, highlighting the specificity of the interactions observed. Figure 2B, equimolar amounts of histones H2A, H2B, H3, and We conclude that RAG1 interacts with the N-terminal tail of H4 were separated by 16% SDS-PAGE, transferred to nitrocellu- histone H3. lose membrane, and probed using purified GST or GST-RAG1FL (left panel). Proteins bound to membrane-associated histones N-Terminal RAG1 Contains a RING Domain were detected by western blotting using anti-GST antibody (right that Mediates Monoubiquitylation of Histone H3 panel). GST-RAG1FL bound most strongly to H3 (lane 1) with In Vitro and In Vivo some weak interaction with H4 being observed in Figure 2B. The N-terminal region of RAG1 contains a RING finger domain As RAG1 consistently displayed preferential binding to histone (aa 264–389, Figure 3A) and has been shown to act as an E3 H3, we therefore characterized the interaction further and ubiquitin ligase that mediates ubiquitylation in vitro (Yurchenko explored a possible functional relevance to V(D)J activity in vivo. et al., 2003) and RAG1 autoubiquitylation in vivo (Jones and GelCore histones are schematically composed of a globular lert, 2003). The above results showed that the N-terminal domain domain that is folded into the nucleosome core and an of RAG1 (1–383) binds histones. Therefore, we investigated N-terminal tail that provides a surface for recognition by whether this domain may also catalyze ubiquitylation of the complexes that regulate chromatin structure and function. To histone H3 to which it binds. To address this issue, we performed map the region of histone H3 responsible for binding to RAG1, an in vitro ubiquitylation assay. A preparation of FLAG-tagged we prepared recombinant H3 globular domain (tailless histone RAG1 and FLAG-tagged RAG2, immunopurified from mammaH3). As shown in Figure 2C, full-length (lanes 1 and 3) and tailless lian cells, was incubated with free ubiquitin (Ub) and recombinant



1

290 328

383

Figure 3. RAG1 Mediates Histone H3 Ubiquitylation In Vitro and In Vivo

1008 1040

A

core N-terminal

307*

325*

328

5G C3 2

-R AG 1 FL

-R AG 1

FL

FL

FL

FL

-R

-R AG 1

D

AG 1

B

H3 0

7A

290

E1 + E2

E1 + E2 ubH3

25

ubH3

25

WB: -H3

WB: -H3

19

19

H3

-R

AG 1 -R FL

-R FL

E

FL

FL

FL

-R

AG 1

w

AG 1

t

w

t w AG 1 -R FL

FL

C

H3 AG 07 A 1 C3 25 G

1 2 3 4 5 6

FL

3 4 5

t

1 2

H3

ubH3

ubH3

(A) Schematic representation of RAG1 domains. The RAG1 N-terminal region (aa 1–383) contains a C3HC4 RING finger (R) motif (aa 290–328) closely associated with an adjacent C2H2 Zn2+ finger (Z) motif (aa 349–389). The entire amino acid sequence of the RING finger motif is also shown, and two point mutations used in this study (H307A and C325G) are marked by asterisks. (B and D) In vitro ubiquitylation of H3. FLAGRAG1 and FLAG-RAG2 proteins were incubated with ubiquitin (Ub), recombinant histone H3, and the indicated combination (+/ ) of mammalian E1 and recombinant E2 (ubcH5b) proteins. Appearance of slower-migrating monoubiquitylated histone H3 forms (indicated by an arrow) was detected by immunoblotting with anti-H3 antibody. (C and E) In vivo ubiquitylation of H3. Expression plasmids, as indicated on the top of each panel, were cotranfected with a plasmid encoding HA-ubiquitin into 293 cells. Plasmid encoding RAG2 was cotransfected to allow RAG complex formation. (C) After immunoprecipitation, proteins were analyzed by western blotting using anti-H3 antibody (left) or anti HA-antibody (right). (E) Histones were isolated from transfected cells, separated by 16% SDS-PAGE electrophoresis, and immunoblotted with an anti-H3 antibody (bottom). Protein levels were detected by immunoblotting with anti-FLAG antibody (top). Monoubiquitylated H3 forms (Ub-H3) are indicated by an arrow. Ig light chains (LC) are also indicated. Markers are expressed in kilodaltons.

WB: -FLAG 25

25

LC

LC

ubH3

with the size of the monoubiquitylated form of histone H3. The same band was H3 H3 visualized after western blot analysis using an anti-HA antibody (Figure 3C, 1 2 1 2 3 4 1 2 lane 2, right panel). Expression of FLAG WB: α- H3 WB: α- ub WB: α- H3 alone did not induce H3 ubiquitylation (lanes 1). The result suggests that ectophistone H3 in the presence of purified mammalian E1 and ically expressed RAG1 is able to increase the level of endogerecombinant E2 (UbcH5B) enzymes. As shown in Figure 3B, nous monoubiquitylated histone H3. western blotting analysis using an anti-H3 antibody detected a band corresponding to monoubiquitylated histone H3 under Mutations in the RING Domain of RAG1 Reduce these conditions (lane 5), but not when RAG1/2 proteins (lanes Monoubiquitylation of Histone H3 1–3) or E1 and E2 (lane 4) were omitted. This result demonstrates Jones and Gellert (2003) showed that an intact RING finger that histone H3 is a substrate for the E3 ubiquitin ligase activity domain was necessary for RAG1 E3 ligase activity; a point mutaof RAG1 in vitro. To explore the physiological relevance of tion introduced at the histidine residue (H307A) displayed lower RAG1-mediated histone ubiquitylation, we determined whether RAG1 E3 ligase activity, as did a point mutation at one of the endogenous histone H3 is ubiquitylated by RAG1 in vivo. Con- conserved cysteine residues (C325G) that coordinates Zn2+ structs encoding FLAG-RAG1 and HA-ubiquitin were cotrans- ions in the RING domain. Of note, a naturally occurring RAG1 fected into 293 cells. A FLAG RAG2 construct was also included mutation in the RING domain (C328G in human corresponding to allow RAG1/RAG2 complex formation. Cell extract derived to amino acid 325 in mouse) is associated with a fatal primary from transfected cells was subjected to immunoprecipitation immunodeficiency called Omenn’s syndrome, in which V(D)J with anti-H3 antibody. Western blot analysis with anti-H3 anti- recombination is impaired (Simkus et al., 2007). Moreover, two body shows that overexpression of RAG1 (Figure 3C, lane 2, N-terminal RAG1 truncation mutants have been found to be left panel) resulted in a slower-migrating band that correlates defective in B cell development while supporting limited T cell 19

19

Molecular Cell 37, 282–293, January 29, 2010 ª2010 Elsevier Inc. 285


development (Villa et al., 2001). Therefore, we tested whether the RAG1 H307A and C325G mutations affect the efficiency of RAG1-mediated histone H3 monoubiquitylation. As shown in Figure 3D, the indicated RAG1 proteins were tested in an in vitro ubiquitylation assay. Reaction products were resolved by 15% SDS-PAGE, transferred to membrane, and probed with anti-H3 antibody. We detected monoubiquitylated histone H3 forms only when all the ubiquitin-ligase reaction components were present (Figure 3D, lanes 2, 4, and 6). RAG1 mutants (RAG1 H307A, lanes 3 and 4; RAG1 C325G, lanes 5 and 6) displayed substantially decreased efficiency in histone H3 monoubiquitylation activity when compared with RAG1 wild-type (lanes 1 and 2). Note that as the mutations represent single amino acid substitutions outside the catalytic domain, we would not expect to fully abrogate RAG1 ubiquitin ligase activity. We conclude therefore that an intact RING domain is required for efficient RAG1-mediated histone H3 monoubiquitylation in vitro. We next measured the effect of these mutants in vivo on RAG1-mediated endogenous histone H3 ubiquitylation. We cotransfected human 293 cells with plasmids encoding FLAG-RAG1 wild-type, FLAG(H307A) RAG1, or FLAG-(C325G) RAG1 together with a plasmid encoding HA-ubiquitin. Histones were extracted from transfected cells and separated on a 16% polyacrylamide gel. As shown in Figure 3E, western blot analysis using an anti-H3 antibody revealed that, in the absence of transfected FLAG-tagged RAG1, cells contained a low level of RAG1-independent monoubiquitylated histone H3 (lane 1) that was substantially increased in the presence of ectopically expressed RAG1 (lane 2). The ectopically expressed WT and mutant proteins are expressed at similar levels (see Figures 5B and 5C). The increased H3 ubiquitylation we see on overexpression of RAG1 most likely includes modification of non-RSS-associated nucleosomes. However, this does not detract from the conclusion of this experiment that RAG1 ubiquitylates endogenous histone H3. Consistent with their reduced activity in vitro (Figure 3D), expression of either of the point mutants, FLAG-RAG1 H307A (Figure 3E, lane 3) or C325G (lane 4), led to reduced levels of monoubiquitylated histone H3 compared to the WT protein (lane 2). A marginal increase in the background level of H3 monoubiquitylation was observed in the presence of the ectopically expressed RAG1 mutants, owing to their residual ubiquitylation activity as detected in vitro. The same monoubiquitylated H3 pattern was observed by western blot analysis using anti-HA antibody (data not shown). The RING domain RAG1 mutants displayed a 40%–60% reduction in ubiquitylation activity (Figure 3 and Figure S2). Of note, N-terminal truncated RAG1 proteins in Omenn’s syndrome, a severe autosomal recessive primary immunodeficiency, retain partial V(D)J recombination activity. The subset of Omenn’s patients with antigen receptor repertoires of limited clonality (Santagata et al., 2000) provides genetic evidence consistent with the assumption that the N-terminal region of RAG1, containing the RING domain, is dispensable for limited recombination activity but that efficient V(D)J recombination requires an intact N-terminal domain. Taken together, the above in vitro and in vivo results strongly suggest that an intact RING domain is required for RAG1-mediated histone H3 monoubiquitylation. RING domain mutations are outside the catalytic core

region of RAG1, but they could be critical for the correct folding of the functional protein. However, we confirmed that separate RAG1 functions are retained by the mutants, indicating that lower ubiquitylation activity of mutant RAG1 is not explained by a failure to form a biologically active structure; mutations in the RING domain do not reduce RAG1 endonuclease activity in vitro and do not disrupt histone binding by the amino tail of RAG1 (Figure S3), indicating that the overall folding of the protein is not unduly affected by the mutations. Mutations in the RING Domain of RAG1 Impair V(D)J Recombination In Vivo As shown above, RAG1 can promote histone H3 monoubiquitylation in vitro and in vivo in a manner dependent on its RING domain. However, RAG1 endonuclease activity on V(D)J DNA substrates does not require an intact RING domain. Two distinct activities coexist in the same protein: ubiquitylation activity in the RAG1 noncore region and an endonuclease activity in the RAG1 core region. Is there a role for RAG1-promoted histone ubiquitylation in modulating chromosomal V(D)J recombination? To address this issue, we first tested RAG1 RING domain mutants in a standard extrachromosomal V(D)J recombination assay (Hesse et al., 1987). In this system, a V(D)J recombination plasmid substrate is transiently transfected into 293 cells and used as substrate for ectopically expressed RAG proteins. To test the influence of the RAG1 RING mutation on the efficiency of signal-joint formation, the plasmid pJH200 (Hesse et al., 1987) was used as a V(D)J substrate. This plasmid contains a polyoma replication origin enabling its efficient chromatinization in vivo, and forms a signal junction following RAG-mediated deletion of the 200 bp sequence between the 12- and 23-RSS. We confirmed that pJH200 plasmid acquires a chromatin structure after transfection into 293 cells by performing a native chromatin immunoprecipitation (ChIP) analysis. Sonicated chromatin from nuclei harvested 24 hr after pJH200 transfection was immunoprecipitated with anti-H3 antibody. DNA was then extracted from immunoprecipitated samples and PCR analysis performed using oligonucleotides that amplify a region in the pJH200 plasmid immediately upstream of the 12RSS (Figure 4). Histone H3 is associated with pJH200 at or immediately adjacent to the RSS, as indicated by the PCR product derived from the anti-H3 immunoprecipitated sample (lane 5). Similar results were obtained using oligonucleotides amplifying different regions of pJH200, confirming that pJH200 becomes chromatinized during transient transfection (data not shown). Chromatinization of the replication-competent episomal V(D)J substrate is consistent with previous observations that the RSS can act as a nucleosome-positioning element (Baumann et al., 2003). To measure recombination activity of the RAG1 mutants, we cotransfected pJH200 together with constructs expressing RAG2 and the indicated RAG1 sequences (Figure 4B). Fortyeight hours later, episomal DNA was prepared and frequencies of recombination (%) were determined using the bacterial colony counting method (Experimental Procedures). Compared with wild-type RAG1 (2.39% recombined), recombination frequency decreased when RAG1 H307A (0.47% recombined), RAG1 C325G (0.45% recombined), or RAG1 C325G/H307A (0.21% recombined) mutants were transfected (Figure 4B). Using PCR



analysis, the efficiency of coding joint formation on transfected substrates was also reduced by mutations in the RING domain of RAG1 (see Figure 5, below). We conclude that mutations in the RING domain of RAG1, although not disrupting DNA cleavage activity, compromise V(D)J recombination in vivo on episomal chromatin substrates.

A

pJH200 (7440 bp)

1

2

3

4

+

-D

5% 2.5% 0.5% -

NA

α-H3 Ip

input DNA

5

6

B

3.5 3.0 2.5 2.0 1.5

p