1.3.4.1 Partner Of R acl (PO RI)
This 34kDa protein was identified as a Rac partner in the yeast two-hybrid system (Van Aelst et al., 1996). P O R I interacts with Rac in a G T P -d ep en d en t m anner, but not w ith Rho or Cdc42. O verexpressed P O R I synergizes with V 12R as to cause m em brane ruffling but has no effect when expressed alone or with V I 2 Rac. A truncation of P O R I that does not bind Rac in h ib its V 12 R ac-in d u ced ru fflin g in N IH 3T 3 cells, w hile the F37A L61 Rac mutant which does not induce ruffling also fails to bind full-length P O R I. These data are consistent with a role for PO R I in Rac-induced ruffling.
PO R I also interacts with the ARF-related protein A R F 6 . A PO R I truncation that does not bind A R F6 inhibits A R F6-induced actin polym erisation at the cell periphery (D 'S o u za-S ch o rey e t al., 1997). This implicates PO RI in cytoskeletal control by A R F6 as well as Rac. P O R I has no kn o w n en zy m atic a ctiv ity or conserved motifs (apart from a putative leucine zipper) and its biochem ical role in cytoskeletal rearrangem ents is unclear.
1.3.4.2 p21-Activated Kinase (PAK)
p 6 5 ^ a serine/threonine kinase related to the budding yeast protein Ste20, was the first target to be identified for Rac. It was isolated using a ligand overlay assay by M anser et al (M an ser et al., 1994). There are now three m am m alian PA K proteins termed P A K l (also known as P A K a ), PAK2 (also known
as PAKy or hPAK65), and PAK3 (also known as PAKP) (Fanger et
al., 1997a) with m olecular weights of, respectively, 6 8 , 62 and 65 kDa. PAK2 is ubiquitous, while P A K l is found in the brain, m uscle and spleen and PAK3 is expressed predom inantly in the brain. U pon binding to either activated Rac or C dc42, PA K a u t o p h o p h o r y l a t e s an d its k i n a s e a c t i v i t y is a c t i v a t e d approxim ately 50-fold using M yelin Basic Protein (M BP) as a su b s tra te . In a d d itio n to th e ir C - te r m in a l s e r in e /th r e o n in e kinase dom ain, PAKs have p otential SH 3 -b in d in g m otifs. The adaptor m olecule Nek, through its second SH3 domain, has been shown to interact with PAK, resulting in PA K relocalisation to
the plasma membrane and activation of its kinase activity (Lu e t a l , 1997).
Since Ste20 in yeast is known to regulate the pherom one response M AP Kinase-related pathw ay (see below), a num ber of groups have investigated the possibility that PA K m ediates Rac- in d u ce d activ a tio n o f the JN K /S A P K and p38 M A P k in ase c a sc a d e s in m a m m a lia n cells. It h as b e e n r e p o r te d th at overexpression of activated PA K in Cos-1 or Cos-7 cells leads to activation of both the JNK and p38 cascades (Zhang et a l , 1995, B row n et a l , 1996, Bagrodia et a l , 1995, Frost et a l , 1996). However, according to another report, co-expression o f w ildtype PAK diminished rather than increased Rac or Cdc42-induced JNK a ctiv a tio n in Cos-7 cells. T hus, w ild ty p e P A K acted as a d o m in an t neg ativ e (T eram oto et a l , 1996a). M oreover, a 43D Rac mutant, which was unable to activate PAK, was still able to induce JNK activation (Westwick et a l , 1997).
A recent report show ed that, while activated PA K (L83, L 86) stimulated both the JNK and p38 cascades in Rat-1 cells, a triple m utant (L83, L 86, R299), which is dead for both GTPase binding and kinase activity did not have a dom in an t negative effect upon Rac- or Cdc42-induced JNK or p38 activation (Tang et a l , 1997). H ow ever, this triple m u tan t b lo ck ed o n co g en ic tran sfo rm atio n of R a ti cells induced by K i-R as and p artially blocked K i-R as-induced (but not v-Raf-induced) E R K activation. The authors p ro p o se d th at PA K has a role in R a s-in d u ce d transformation, possibly through the E R K cascade. In support of this idea, activ ated PAK, Rac and C dc42 w ere re p o rte d to synergise with c-R afl for activation of the ERK M A PK pathw ay (Frost et a l , 1997, Tang et a l , 1997). F u rth e rm o re , a ctiv a ted PAK was able to directly phosphorylate MEK-1 on serine 298 in vitro. This p h o sp h o ry la tio n did not in cre ase M E K l 's k in ase activity, but appeared to increase its affinity for its u pstream regulator, R af (Frost et a l , 1997). It is therefore possible that, via PAK, Rac and Cdc42 could control ERK activity in a "cross cascade" fashion. The role of PAK in m ediating Ras-, Rac- and C dc42-induced M AP kinase pathway activation is still not clear, though it could be that PA K is resp o n sib le for this G T Pase function in some cell types but not others.
A num ber of proteins bearing hom ology to PA K such as H P K l (H em atopoietic Progenitor Kinase), NIK (Nek Interacting Kinase) or GCK (Germinal C enter Kinase) have been show n to activate JN K when overexpressed (Su et al., 1997, Fanger et al., 1997a). H P K l binds to MLK3 and NIK binds to M E K K l; these kinases could therefore d irectly be connected to JN K, though none have so far been shown to bind Rac/Cdc42.
Some reports have suggested a link betw een PA K and the cytoskeleton. One group claims that kinase-dead PA K (L83, L 86, R 299) leads to form ation o f filo p o d ia and m em b ran e ru ffles w hen o v erex p ressed in Swiss 3T3 cells (Sells et al., 1997). H o w ev er, o th er groups re p o rte d that m u ta tio n s o f R ac and
Cdc42 at codon 40, which abolish binding to PAK, did not affect
Rac- or C d c4 2 -m ed iated c y to sk eleta l rea rran g e m en ts in Swiss 3T3 cells (Lamarche et al., 1996, Joneson et al., 1996, W estw ick et al., 1997). A nother group dem onstrated th at in tro d u ctio n of activated PAK in HeLa cells resulted in loss o f stress fibres and focal adhesions, and proposed that part o f PAK's function is to antagonise R ho (M anser et al., 1997). M a m m a lia n PAK, like
Ste20 in yeast, was also shown to phosphorylate the M yosin I
heavy chain, leading to an in cre ase in its A T P ase activ ity (B rzeska et al., 1997). Despite all these studies, the precise role of PAK in cytoskeletal regulation remains confusing and unclear.
A recent development in the study o f PA K function is the finding that PAK2 is proteolytically cleaved by caspases during F as-m ed iated apoptosis in Ju rk at T -cells (R udel and B okoch,
1997). This event yields a 34 kD a C-terminal fragm ent which is c a ta ly tic a lly a ctiv e. F u rth e rm o re , the a u th o rs sh o w e d th a t overexpression of dom inant negative PA K blocked some o f the m o r p h o l o g i c a l c h a n g e s th a t o c c u r d u r i n g F a s - m e d i a t e d apoptosis. This raises the possibility that PA K is an im portant player in the apoptotic program m e. A nother rep o rt claim ed that X enopus PA K protected G 2 -arrested O ocytes ag ain st apoptosis (F au re et al., 1997). T h ese two fin d in g s are n o t m u tu a lly exclusive, since it is thought that m any proteins can both cause and p r e v e n t a p o p to s is d e p e n d in g on the c e llu la r c o n te x t
(Baichwal and Baeuerle, 1997, Herdegen et al., 1997).
1.3.4.3 Mixed Lineage Kinase (MLK)
By taking the minim al Rac/Cdc42-binding dom ain of PAK and searching protein databases, B urbelo et al id en tified a 16 amino acid m o tif com m on to a large n u m b er o f R ac/C d c4 2 - bin d in g p ro tein s (B urbelo et al., 1995). This so -called CRIB (Cdc42/Rac Interactive Binding) m otif is found on M L K 1/2/3, a
family of kinases similar to MAPKKKs (Dorow et al., 1993, Katoh
et al., 1995). In addition, MLKs have a serine/threonine kinase dom ain, an SH3 dom ain, a leucine zip p er and a C -term in al divergent region. MLK2/3 were found to bind Rac and Cdc42 in a G TP-dependent m anner and both are potent activators o f the
JN K and p38 M A P K inase path w ay s (B urbelo et al., 1995,
T eram o to et al., 1996a, Tibbies et al., 1996, Ran a et al., 1996, Nagata et al. 1998, Hirai et al., 1997). MLK2/3 have been shown to activate JN K via SEK-1 and M LK3 was d e m o n stra te d to activate p38 via MKK3/6 (Tibbies et al., 1996, Hirai et al., 1997). MLKs appear to be constitutively active in transfection assays; how they are regulated in vivo is not clear.
N agata et al recently reported that the M LK2/3 divergent region associates with 14.3.3e, a calcium -binding protein called hippocalcin, two kinesin-related proteins KIF3X and KIF3A and a K IF 3-binding protein, K A P3A (N agata et al. 1998). In these studies, M L K 2 co -lo c alised with K IF 3 A along m ic ro tu b u le s, suggesting that the kinase m ight p lay a role in m ic ro tu b u le motor function. Rac has been reported to bind to tubulin directly in vivo, though the physiological significance of this interaction,
if any, is unknown (Best et al., 1996). M SE55, ano th er CRIB
containing protein, has no known function (Burbelo et al., 1995).
1.3.4.4 M E K K l/4
M EKK 4 is a CRIB containing m em b er of the M A PK K K fa m ily . B o th M E K K l and 4 h a v e b e e n s h o w n to co- imm unoprecipitate from cell lysates with Rac and C dc42 and to
bind directly to these GTPases in vitro (Fanger et al., 1997b).
M E K K l and 4 both activate the JNK pathway, but only M E K K l is able to give som e E RK activation and n eith er activ ates p38 (Lange-C arter et al., 1993, Gerwins et al., 1997). M E K K l is a 196 kD a pro tein with a p ro lin e-rich m o tif and two PH dom ains.
M E K K l has been shown to bind Ras in a GTP-dependent manner (Gerwins et al., 1997). MBKK4 is 180 kDa, similar to M E K K l, but has only one PH domain, a proline-rich region, and a CRIB m otif (Gerwins et a l , 1997). Both o f these proteins are good candidate effectors for Rac and Cdc42, but not Rho, in activating JNK.
1.3.4.5 S6 Kinase
There has been one report that p p 7 QS6 K binds to and is activated by GTP-bound Rac and Cdc42, though it is not clear whether the interaction is direct or not (Chou and Blenis, 1996). p p 7 0 ^ ^ ^ p lays an esse n tia l p a rt in G l - S tra n s itio n since m ic r o in je c tio n o f n e u tr a lis in g a n tib o d ie s a g a in s t p p 7 0 ^ 6 K i n h i b i t s s e r u m - i n d u c e d D N A s y n t h e s i s ( P r o u d , 1 9 9 6 ). F u rth e rm o re , p p 7 0 ^^K activated by m any m itogenic stimuli, including growth factors, cytokines, phorbol esters as well as by o n c o g e n e s (C h o u an d B le n is , 1995). D o m in a n t n e g a tiv e Rac/C dc42 inhibited activation of p p 7 0 S6K by serum (Chou and B le n is, 1996). p p 7 0 S 6 K th erefo re a c an d id ate targ e t for Rac/Cdc42 in mediating cell cycle progression through G l.
1.3.4.6 PRK2 and Citron
These are two Rho targets (see above) that have also been shown to bind Rac, though the p h ysiologial relev an ce o f this binding rem ains to be pro v en (V in cen t and Settlem an, 1997; M adaule et al., 1995).
1.3.4.8 PI4P5K
It has been rep o rted th at th ro m b in and a c tiv a te d Rac m ediate uncapping and severing o f actin filam ents in platelets (H a rtw ig , 1995). T his is in a g re e m e n t w ith the w o rk o f M a c h e s k y and H a ll w h o fo u n d th e R ac in d u c e d ra p id incorporation of fluorescent actin m onom ers at areas o f lam ella form ation (M achesky and Hall, 1997). Furtherm ore, H artw ig et al were able to inhibit actin filam en t uncap p in g using PIP2- binding peptides supporting the involvem ent o f a PI4P 5K as a m ediator o f R ac-induced actin polym erisation. Indeed, T olias et al purified a PI4P5K activity by affinity chro m ato g rap h y with G T P/G D P-bound Rac, though the identity and function o f the
enzym e rem ains to be investigated (Tolias, 1995). Rac was also shown to bind PI3K, but the significance o f this fin d in g is unknow n (Tolias, 1995).