31/-400
Bait: RhoA mutants PKN positive N egatives selected 36/107 Target: PKN (fused to GAL4 activation domain) Target: ROCK (fused to GAL4 activation domain) RhoA mutants PKN positive ROCK negative
Mutations
C20R/T37Y, C20R/F39L, E40L and E40W
Figure 2.2A: Screening of RhoA mutant libraries. Outline of screening procedure for selection of PKN positive, ROCK negative RhoA mutants. Number of transformants screened and selected and mutations recovered are shown in larger type.
Bait: RhoA mutants (fused to GAL4 DBD)
Positives selected
68/402
Bait: RhoA mutants ROCK positive N egatives selected Target: ROCK (fused to GAL4 activation domain) Target: PKN (fused to GAL4 activation domain) 16/84 RhoA mutants ROCK positive PKN negative
Mwtations
F39V, E40N and E40T
Figure 2.2B. Screening of RhoA mutant libraries. Outline of screening procedure for selection of ROCK positive, PKN negative RhoA mutants. Number of transformants screened and selected and mutations recovered are shown in larger type.
2.2.4 Screen for RhoA m utants able to bind ROCK but not PKN
The screen to isolate RhoA m utants able to bind ROCK but not PKN w as similar the screen to isolate RhoA m utants able to bind PKN b u t not ROCK (described in Figure 2.2B). One m icrogram of each of the four GAL4DBD-RhoA m u tan t libraries w as tran sfo rm ed into HF7C yeast carrying the GAL4AD-ROCK construct. 402 transform ants w ere screened for lacZ rep o rter expression (approxim ately one hundred from each position m utant library). Plasm id DNA w as rescued from sixty eight transform ants which produced a strong blue colour w ith in three hours w hen assayed in a colony colour assay, indicating a strong interaction w ith ROCK. As before, the plasm id DNA w as digested w ith Clal to linearise the GAL4AD-ROCK construct prior to bacterial transform ation. The pool of ROCK positive GAL4DBD-RhoA m utant plasm ids was transform ed into HF7C yeast carrying GAL4AD-PKN and the transform ants screened for lack of interaction w ith GAL4AD-PKN. Plasmid DNA w as recovered from sixteen of the eighty four transform ants screened w hich failed to grow on plates lacking histidine, indicating a lack of interaction w ith PKN. These ROCK positive/PK N negative GAL4DBD-RhoA m utants were checked by retransform ation into HF7C yeast w ith the GAL4AD-ROCK and GAL4AD-PKN constructs. Sequencing revealed three m utants, F39V, E40N and E40T.
2.2.5 Other RhoA effector loop m utants
In addition to the RhoA m utants isolated in the screens outlined above I m ade two other effector loop m utants based on previously published studies of Racl
and Cdc42. Lamarche et al dem onstrated that m utations F37A and Y40C in Racl
and Cdc42 disrupted binding to PAKl and ROCK, respectively, in the yeast two- hybrid system (Lamarche et a l 1996). I m ade the corresponding m utations in RhoA: F39A and Y42C. I have also tested the effector interactions of the C3 transferase resistant RhoA m utant, N41I (Hill et al. 1995).
2.3 Interaction of RhoA m utants w ith a panel of effectors
2.3.1 Two hybrid analysis of the binding properties of RhoA m utants
The yeast tw o-hybrid assay w as used to analysis the interaction of the RhoA m utants isolated in the screens described above w ith mDia2, citron, rhophilin, kinectin, and m NETl as well as PKN and ROCK. Four m utants w ere recovered in the screen for RhoAV14 derivatives that can interact w ith PKN b u t not ROCK; C20R/T37Y, C20R/F39L, E40L and E40W. M utant C20R/F39L show ed greatly im paired interaction w ith ROCK, Rhophilin, Kinectin and Citron (Table 2.2, line
Gal4-DBD fusion His3 activity ^ with Gal4-AD fusion None PKN 1-942 PKN 1-511 ROCK 348-1018 Rhophilin 1-130 Kinectin 1053-1327 Citron 647-780 mDia2 47-800 mNETl 1-596 RhoAV14-WT-S190 1 2 4 >4 3 >4 >4 3 T37Y/C20R<^ (1) (1) (1) (1) 2 (1) (1) (1) 3 F39A^ - - - >4 - F39V^ - - - 4 - - - >4 4 F39L/C20RC - 1 2 1 1 - - >4 3 E40L^^ (1) (1) 2 (1) 4 (1) >4 >4 (1) E40W(: (1) (1) 2 (1) >4 (1) 4 >4 (1) E40N^ - - 2 4 >4 1 >4 >4 3 E40T^ - - 2 4 >4 - >4 >4 3 N 4lM - - - 4 - - >4 >4 nd Y42C^ . 4 2 3 >4 >4 2
Table 2.2: Interaction of RhoA effector loop m utants w ith effectors. Growth of cells carrying the indicated plasm ids was evaluated by a
sem iquantitative plate assay for HIS3 activity w as by grow th on plates containing increaseing am ounts of 3-aminotriazole. (a) Grow th
was scored as follows: (-) no grow th on plates lacking histidine; (l->4) grow th on plates containing OmM, Im M , 2mM, 4mM, 8mM, more
than 8mM am inotrizole respectively. Scores in parenthesis indicate that these RhoA m utations prom oted grow th on plates lacking
histidine in the absence of an effector plasm id, (b) M utants recovered from the ROCK+ /PK N - screen, (c) M utants isolated from PKN+
5) b u t w ild type interactions w ith PKN, mDia2 and m N E Tl. The other three m utants, C20R/T37Y, E40L, and E40W (Table 2.2, lines 2,6, & 7), exhibited an increased background activity in the assay in the absence of the activator-tagged effector, w hich did not increase in the presence of the full length PKN; nevertheless, m u tan ts E40L and E40W in teracted in d istin g u ish ab ly from RhoAV14S190 w ith the N -term inal region of PKN (Table 2.2, colum ns A-C; see Discussion). The E40L and E40W m utations reduced interaction w ith ROCK(348- 1018), kinectin and m N ETl to background levels, b u t other interactions were essentially unim paired (Table 2.2 colum ns D, E, F, G, H & I). The fourth m utation recovered in the PKN positive/R O C K negative screen, C20R/T37Y, show ed no interaction above background level w ith any of the effectors tested ap art from R hophilin, w hose interaction w as greatly red u ced relative to w ildtype, and m N ETl (Table 2.2, line 2); this behaviour is discussed further below.
The screen for RhoA m utants able to bind ROCK but not PKN isolated m utants. F39V, E40N and E40T. Of these, m utant F39V exhibited profoundly im paired interaction w ith all effectors tested apart from ROCK and m N ETl (Table 2.2, line 4). In contrast, m utants E40N and E40T exhibited su b stan tially different behaviour. A lthough these m utants were selected for their inability to bind PKN, they interacted effectively w ith its N-terminal dom ain (Table 2.2: colum ns B & C; see Discussion 7.1 & 7.2). Their interaction w ith other p roteins w as largely unim paired, apart from kinectin, w hich w as substantially reduced (Table 2.2: lines 8,9). In addition, I tested the binding properties of the RhoA m utants F39A and Y42C, which are based m utations previously m ade in Racl and Cdc42, and m utant N41I, which is resistant to C3 toxin (Hill et al. 1995; Lamarche et al. 1996). M utations F39A, N41I and Y42C prevented interaction w ith PKN(1-511) and intact PKN. M utation E39A disrupted interaction w ith all other effectors tested apart from mDia2(47-800) (Table 2.2: line 3), while m utations N41I and Y42C did not affect the interactions w ith the other effectors tested except rhophilin (Table
2.2: lines 1 0 & 1 1).
2.3.2 GST-pull dow n analysis of the binding properties of RhoA m utants
To corroborate the two hybrid interaction data I used an in vitro binding assay. GST fusion genes carrying PKN(1-511), ROCK(348-1018) and mDia2(47-257) were bacterially expressed. After purification approxim ately one m icrogram of each GST fusion protein immobilised on glutathione-sepharose beads w as incubated w ith lOOng of the 9E10 epitope-tagged RhoAV14 m utants, w hich h ad been
previously loaded w ith GTP7S. After extensive w ashing the levels of RhoA
protein bound to the imm obilised GST fusion protein w ere m easured by 9E10 74
im m unoblot. The results are show n in Figure 2.3. The binding of the RhoA m utants to the ROCK fusion proteins was very sensitive to the w ash conditions used; lithium chloride was required in the w ash buffer to disrupt the binding of
m utants E40L and E40W. In general the in vitro binding studies corroborated the
tw o-hybrid assays. The ability of the m utants to bind PKN(1-511) reflected the interactions seen in the tw o-hybrid assay, although w eak interactions w ere detectable w ith m utants C20R/T37Y, F39V and Y42C upon prolonged exposure (see Figure 2.3B). Effector loop m utants that interacted efficiently w ith ROCK in the tw o-hybrid assay also bound it strongly in vitro; however, in contrast to the tw o -h y b rid data, m u tan ts C20R/T37Y and E40W also b o u n d w eakly to ROCK(831-1011) in vitro (Figure 2.3C; see Discussion). All the RhoA m utants b o u n d to mDia2(47-257) in vitro (Figure 2.3D). The dram atically different
behaviour of m utant C20R/T37Y in the two hybrid assay and the in vitro binding
assay m ay be reflect the difficulty of this m u tan t to bind GTP in vivo (see Discussion 7.1.1). Taken together w ith the two-hybrid results, these data show that RhoA m utants isolated in the screens described above and the m utants analogous to those in Racl and Cdc42 have selectively lost the ability to interact w ith specific RhoA target molecules.
Effector loop mutation: B D Input Bound to GST- PKN (1-511) Bound to GST- ROCK (348-1018) Bound to GST- mDia2 (47-257) 1 2 3 4 5 6 7 8 9 10 - '' ' 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Figure 2.3: In vitro in teraction s b e tw e e n R ho m utants and effectors. lOOng of GTPyS-loaded bacterially produced 9E10 epitope tagged RhoA m utants (shown panel A ) were incubated with the indicated bacterially produced GST-effectors proteins (lOOOng) imm obilized on glutathione-sepharose
beads. Mutant Rho A VI4 proteins bound to GST-PKNl-511 (shown panel B),
GST-ROCK348-1018 (shown panel C) or GST-mDia247-257 (shown panel D )
were detected w ith 9E10 antibody following , elution, SDS-PAGE and immunoblotting.
3 FUNCTIONAL ANALYSIS OF RhoA EFFECTOR MUTANTS, ROCK AND PKN
3.1 Chapter sum m ary
The functional characterisation of the RhoA effector m utants detailed in Chapter 2 is described in this chapter. The RhoA effector m utants w ere tested for their ability to regulate actin stress fibre formation, transcription by SRF, and loss of contact inhibition. The functional properties of the RhoA m utants have been correlated w ith their binding properties to give an insight into the role of various effectors in m ediating the various functions of RhoA. The role of PKN and ROCK as RhoA effectors has also been tested by overex p ressin g either constitutively active or inactive versions of the proteins.
3.2 Regulation of the cytoskeleton by RhoA and its effectors
3.2.1 Regulation of the cytoskeleton by RhoA effector m utants
The effects of the effector loop m utations described in C hapter 2 on RhoA- in d u ce d cytoskeletal reo rg an isatio n w ere in v estig ated by m icroinjecting expression plasm ids encoding 9E10 epitope-tagged RhoAV14 or effector loop m utants into serum -deprived NIH3T3 cells; filam entous actin and RhoA w ere visualised by staining w ith phalloidin and 9E10 antibody, respectively. Serum- starved NIH3T3 cell have low levels of filamentous actin, usually organised into a few short fibres w ith no particular organisation w ith respect to one another
(Figure 3.1A top left panel). E xpression of RhoAV14 p ro m o te d the
reorganisation of filamentous actin into parallel arrays of stress fibres five hours after plasm id microinjection (Figure 3.1 A top m id panel). I also exam ined focal adhesion complex form ation by staining w ith vinculin antibodies; how ever, the relatively high basal level of focal adhesions in serum -deprived NIH3T3 cells did not increase upon expression of RhoAV14 (Figure 3.1C). This prevented analysis of focal adhesion contact formation by the RhoA effector loop m utants.
All the RhoA m utants were expressed at similar high levels and show ed similar partly punctate cytoplasmic distribution (Figure 3.1 A inset panels). How ever, they varied in their ability to induce stress fibre form ation. E xpression of m utants E40N, E40T, and Y42C triggered efficient stress fibre formation, though those form ed by E40T w ere shorter and less num erous than those form ed by RhoAV14 (Figure 3.1A&B). M utant N41I also efficiently prom otes stress fibre form ation (Art A lberts & Richard Treism an p erso n al com m unication) .In contrast, expression of RhoAV14 m utants F39A, F39V, C20R/F39L, E40L, and
I -3 ‘) V + K39A I 3 9 V + IW>L I 3 9 V + i ; 4 » \ v Kvn + l* K Niat K39V + K 4 0 I .
F ig u re 3.1A: R e g u la tio n o f th e c y to s k e le to n b y R h oA e ffe c to r m u ta n ts. NIH3T3 cells m aintained in 0.5% serum were microinjected with myc epitope tagged RhoA expression constructs, Rho expressing cells and actin cytoskeleton were visualised using 9E10 antibody (inset panels and marked with arrowheads) and TRITC-phalloidin, respectively (20-30 cells injected per coverslip). Representative cells are shown from one of three independent experiments.
^ 60"
R h oA .V 14 m u tan t
+ F39V
F igu re 3 .IB: R e g u la tio n o f th e c y to s k e le to n b y R h oA e ffe c to r m u ta n ts. Proportion of injected cells with large number of long stress fibres arranged in parallel sheets (average of three independent experiments; error bars indicate SEM).
M ock R h oA V 14
F ig u r e 3.1C: R e g u la tio n o f fo c a l a d h e s io n s b y R h o A . NIH3T3 cells m aintained in 0.5% serum were microinjected w ith a myc epitope tagged RhoAV14 expression construct, focal adhesions were visualised using anti- vinculin antibody (injected cells marked with arrowheads, 20-30 cells injected per coverslip). Representative cells are shown from one of two independent experiments.
E40W did not lead to efficient stress fibre formation (Figure 3.1A&B). The RhoA effector m utants E40N, E40T, N41I and Y42C all prom oted stress fibre form ation and all bound to mDia2, citron and ROCK (see Table 3.1, also C hapter 2.3). This is consistent w ith previous reports that implicate mDia and ROCK/Rho-kinase in re g u la tio n of the actin cytoskeleton by RhoA (see C h a p te r 1.6&1.7). RhoAV14V39 did not trigger stress fibre formation, even though its interactions w ith ROCK and mDia2 were unim paired, suggesting additional effectors m ay be required for stress fibre formation. To test this I investigated w hether the F39V m u ta n t could functionally co-operate w ith the RhoA m utants unable to bind ROCK (F39A, C20R/F39L, F40L & F40W) or an active PKN construct (PKNcat - see Section 3.2.2). Simultaneous expression of m utant F39V w ith F40W or F40L trig g ered stress fibre form ation (Figure 3.1A&B). In contrast, sim ultaneous expression of m utant F39V w ith F39A, C20R/F39L or PKNcat did not lead to changes in the actin cytoskeleton. M utants F40N, F40T, N41I and Y42C dem onstrate that the ability to regulate stress fibre form ation does not require binding to PKN, rhophilin, and kinectin (Table 3.1). Taken together, these data are consistent w ith stress fibre form ation requiring interaction of RhoA w ith RO CK /Rho-kinase, m D ia/m D ia2, and an additional effector(s), possibly citron. Interaction w ith PKN, rhophilin or kinectin is not required (Table 3.1).
3.2.2 Regulation of the actin cytoskeleton by ROCK and PKN
Studies w ith RhoA effector m utants suggest that ROCK bu t not PKN plays a role in stress fibre form ation by RhoA. To test this directly I expressed m u tan t versions of ROCK and PKN. These are detailed in Figure 3.2A together. The kinase activities of the ROCK and PKN m utants were determ ined by transfection into NIH3T3 cells followed by imm uneprecipitation and kinase assays or w estern blotting. The kinase activities of the im m uneprecipitates w as assayed by in cu b atio n w ith y32P-ATP, and m yelin basic p ro tein and m ixed h istone su b strates (Figure 3.2B). The kinase activity of PKN and ROCK w as n o t increased u p o n serum stim ulation and only m odestly by co-transfection of RhoAV14 (Figure 3.2B). The modest activation of ROCK and PKN kinase activity by RhoAV14 precluded testing the regulation of ROCK and PKN kinase activities by the RhoA effector m utants. Truncation of ROCK and PKN to ROCKA3 and PKNcat, respectively, increased the kinase activity im m uneprecipitated; though bo th constructs expressed better than their full length co u n terp arts m aking evaluation of the specific activity of the constructs difficult (Figure 3.2B&C).
Microinjection of ROCK or ROCKA3 expression plasm ids into NIH3T3 cells led to the form ation of thick actin fibres. These fibres w ere thicker, shorter and less num erous than those triggered by RhoAV14 (Figure 3.3). The fibres triggered by
RhoA protein Interaction ^ Activity
ROCK PKN Citron mDia2 Actin ^ SRF ^ Foe
For RhoAV14-WT + + + + + + + + + + + + + + F39A - - - + + - + - F39V + + - - + + - + + + F39L/C20R -/+ + + - + + - + - E40L - + + + + + + - - - E40W -/+ + + + + + + - - - E40N + + + + + + + + + + + E40T + + + + + + + + + + + N41I + + - + + + + + + + + nd Y42C + + _ + + + + + + + + + +
T a b le 3.1: S u m m a r y o f r e s u lts , (a) Level of interaction (++) w ildtype in both tv/o-hybrid and GST pull dow n assays (+) only w ith PKN 1-511 not PKN full length (-/+ ) only detectable one assay, either tw o-hybrid or GST pull dow n (-) no significant interaction in either assay, (b) Stress fibre form ation (++) 50-100% w ildtype activity (+) 20-50% w ildtype activity (-) <20% w ildtype activity, (c) SRF activation (++) w ild type in both assays (+) 50-100% w ildtype activity in microinjection assay - 10-30% activity in transfection assay (-) <50% activity in microinjection assay - <10% activity in transfection assay, (d) Focus form ation (++) 50-100% w ildtype activity (+) 20-50% w ildtype activity (-) <20% activity.
Construct R O C K ROCKA3 RO C K A 3K >A |