(Received for publication, May 31, 1996, and in revised form, October 31, 1996)
From the Department of Cell Biology, Howard Hughes Medical Institute, Yale University School of Medicine, New Haven, Connecticut 06510
Mss4 is a guanine nucleotide exchange factor that specifically binds to, and promotes GDP-GTP exchange on, a subset of the Rab GTPases (Burton, J. L., Burns, M. E., Gatti, E., Augustine, G. J., and De Camilli, P. (1994) EMBO J. 13, 5547-5558). In order to identify the domain(s) of the GTPase that is important for this interaction, protein chimeras were constructed between Rab3a, which binds Mss4, and Rab5a, which does not bind Mss4. We have identified the amino-terminal portion of Rab3a as the Mss4-binding region, with the effector domain being critically required for binding and the flanking regions further enhancing the interaction. Sequence comparisons have revealed that Mss4-binding Rabs share more homology with each other than with Rabs that do not bind Mss4. The region of highest homology between these Rabs, which defines them as members of the same evolutionary branch within the Rab subfamily, coincides with the domain shown here to be critical for Mss4 binding. A mutation in the zinc-binding domain of Mss4 (Mss4 D96H), a region that is highly conserved between Mss4 and its yeast homologue Dss4, completely abolished its property to bind to, and promote GDP-GTP exchange on, Rab3a. Thus, the preservation of the Mss4/Dss4-GTPase interaction appears to have been a critical factor in the evolution of this subset of Rab proteins.
The Rab GTPases, members of the Ras GTPase superfamily, are thought to play a central role in vesicular transport events within the cell. The molecular mechanism utilized by these proteins to control membrane traffic remains unclear. Experimental evidence suggests that their cycling between GDP- and GTP-bound conformations is critical for their function, and accessory proteins that influence the nucleotide state and their membrane-association have been identified (1-3). Such proteins include guanine nucleotide exchange factors (GEFs),1 which stimulate GDP-GTP exchange, GTPase-activating proteins (GAPs), which stimulate GTP-hydrolysis, and guanine dissociation inhibitors (GDIs), which stabilize the GTPase in a GDP-bound conformation and promote membrane dissociation of the GDP-bound form of the molecule (2-5).
Determining the domains of the Rab GTPases that are important for their
subcellular localization and for their interaction with their accessory
proteins may provide clues as to how these GTPases function in membrane
trafficking. The carboxyl terminus or "hypervariable domain" of the
Rabs has been shown to be important for their specific localizations
within the cell (6-8). The "effector domain" (L2/2 according to
Ras nomenclature), together with the L3/
3 domain, the hypervariable
domain, and the geranylgeranylated C terminus has been demonstrated to
be involved in binding to Rab GDI, suggesting that GDI may shield the
hypervariable domain and the lipid moiety of the Rabs, thereby
preventing membrane association (9). Very little is known however,
about the Rab GAPs or GEFs and the domains of the Rabs with which they
interact.
The only two GEFs molecularly identified thus far for Rab GTPases are Mss4 (and its yeast counterpart, Dss4) and the recently identified p619 protein, which appears to stimulate guanine nucleotide release from both Arf1 and some Rab family members (10-12). Mss4 binds to, and stimulates GDP-GTP exchange on, a subset of the Rab GTPases that belong to the same branch of the Ras superfamily (13). To identify the Mss4 binding site on the GTPase, we constructed chimeric molecules between Rab3a, which binds Mss4, and Rab5a, which does not bind Mss4. Using this approach, we have established that the Rab domain necessary for binding to Mss4 coincides with its amino-terminal domain, which defines Mss4-binding proteins as members of the same branch within the Rab family tree. A point mutation in the zinc-binding domain of Mss4 disrupted Mss4 binding to, and GDP-releasing activity from, Rab3a, confirming the critical role of this domain in Mss4 function.
The following reagents were purchased from
commercial sources: 35S-GTPS (1300 Ci/mmol) and
125I-protein A (9.2 µCi/µg) from DuPont NEN;
[35S]sulfur labeling reagent from Amersham Corp.;
restriction enzymes, Vent DNA polymerase, Amylose-Sepharose, and
anti-MBP antibodies from New England Biolabs (Beverly, MA); T4 DNA
ligase (5 units/µl) from Boehringer Mannheim; Sequenase 2.0 sequencing kit from U.S. Biochemical Corp.; pET15b bacterial expression
vector from Novagen (Madison, WI); Ni2+-agarose from Qiagen
(Chatsworth, CA); goat anti-rabbit IgGs conjugated to alkaline
phosphatase from Bio-Rad; nitro blue tetrazolium, 5-bromo-4-chloro-3-indolyl phosphate tetrazolium, and imidazole from
Sigma; Altered Sites II Mutagenesis system from Promega (Madison, WI);
and nitrocellulose circular filters (0.45 µ) from Millipore (Bedford,
MA).
All DNA manipulations were performed using
standard procedures (14). Rab3aC1 and Rab3a
C2 truncation mutants
were made by introducing a premature stop codon in the Rab3a cDNA
at aa positions 218 and 187, by site-directed mutagenesis using the pALTER-Ex1 vector of the Altered Sites II mutagenesis system with the
mutagenic oligonucleotides 5
-CATCAGGATTGAGCCTGCTGA-3
and
5
-ATCTGTGAGAAGTAGTCGGAGTCCCTA-3
, respectively. For chimeras between Rab3a and Rab5a (Rab3/5), the Rab5a cDNA was subcloned into the pALTER-Ex1 vector, and restriction sites were created within Rab5a by site-directed mutagenesis to allow easy insertion of various Rab3a cDNA pieces that were produced by the polymerase chain reaction (PCR) with Vent DNA polymerase. For Rab3/5 I
and III constructs, a Rab5a cDNA was used in which an EcoRI restriction site was removed from the vector using the
oligonucleotide 5
-GGGCGAATTGGTTAACTTTA-3
, and
EcoRI and SacI restriction sites were created
within the Rab5a cDNA by base pair changes at nucleotide position
144 with the oligonucleotide
5
-CAATTTCATGAATTCCAAGAGAGTACC-3
and positions
217, 218, and 219 with the oligonucleotide
5
-AACAGTAAAGTTTGAGCTCTGGGATACAGCTGGT-3
, respectively. Note that the resulting SacI restriction site
within the Rab5a cDNA results in a conservative aa change of
isoleucine to leucine at aa position 73. For Rab3/5 I, DNA encoding aa
51-73 of the Rab3a cDNA replaced the coding DNA for aa 48-72 of
Rab5a by subcloning into the EcoRI and SacI
restriction sites within the Rab5a cDNA. Rab3/5 III was produced by
amplifying the Rab3a cDNA region encoding aa 1-73, which was used
to replace the cDNA region encoding aa 1-72 of Rab5a using the
NcoI and SacI restriction sites. For Rab3/5 II,
IV, V, and VI constructs, a Rab5a cDNA/pALTER-Ex1 construct was
made that, like the Rab5a cDNA above, had a deleted EcoRI site in the vector and a created EcoRI site
in the Rab5a cDNA, but no SacI site. For Rab3/5 II, the
coding region for aa 51-102 of Rab3a was used to replace the region
encoding aa 48-101 of Rab5a using the EcoRI and
EcoRV restriction sites. Rab3/5 IV was made by PCR
amplification of the Rab3a cDNA region encoding aa 1-102, which
replaced the Rab5a cDNA encoding aa 1-101 using the
NcoI and EcoRV restriction sites. For Rab3/5 V,
aa 1-47 of Rab5a was replaced by aa 1-48 of Rab3a by subcloning the
PCR product for this region of Rab3a into the NcoI and
EcoRI restriction sites of the Rab5a cDNA. For Rab3/5
VI, aa 1-47 of Rab5a were replaced by aa 20-48 of Rab3a using the
NcoI and EcoRV restriction sites. Rab3/5 VII was
made by PCR amplification of Rab3/5 III using oligonucleotides that
result in the removal of the DNA encoding the first 29 aa of Rab3a in
the Rab3/5 III construct. Sequence for all of the chimeric constructs
was confirmed by dideoxysequencing. All of the above constructs, as
well as wild-type Rab3a and Rab5a cDNAs, were PCR-amplified using
Vent DNA polymerase with the appropriate oligonucleotides containing
NdeI and BamHI restriction sites for subcloning
into the bacterial expression vector pET15b.
For the Mss4 D96H point mutation, the coding region of Mss4 was
amplified by PCR and subcloned into the pALTER-Ex1 vector using the
XbaI and HindIII restriction sites. The mutation
was introduced using the mutagenic oligonucleotide
5-AATCTCACAGTGTGCACAGAC-3
. The presence of the mutation
was confirmed using dideoxynucleotide sequencing. This construct, as
well as wild-type Mss4, was PCR-amplified using Vent DNA polymerase and
subcloned into NdeI and BamHI restriction sites
of the bacterial expression vector pET15b.
The above
DNA constructs were transformed into BL21 (DE3) bacteria for protein
expression (15). Cells were induced at 30 °C for 4-5 h in the
presence of 0.2 mM
isopropyl-1-thio--D-galactopyranoside and then lysed by
sonication in Buffer A (50 mM Hepes, pH 8.0, 100 mM NaCl, 5 mM MgCl2), 30 mM imidazole, and protease inhibitors. TX-100 was then
added to a final concentration of 0.5%. Insoluble material was
pelleted, and the supernatant was incubated overnight with
Ni2+-agarose. The Ni2+ resin was pelleted and
then washed in sequence with 150 ml of Buffer A plus 0.5% TX-100, 50 ml of Buffer B (50 mM NaPO4, pH 7.2, 300 mM NaCl, 10% glycerol), pH 6.0, and 50 ml of Buffer A and
then eluted with Buffer A containing 250 mM imidazole.
Imidazole was removed by dialysis overnight in Buffer A plus 1 mM dithiothreitol. Proteins produced from constructs Rab3/5
IV and VII were largely insoluble. The insoluble protein fraction was
therefore used in some of the gel overlay experiments after
solubilization in SDS sample buffer. For Mss4 solution binding assays,
the insoluble protein present in inclusion bodies was partially
solubilized in 50 mM Tris, pH 7.5, 8 M urea.
Mss4 fused to the maltose-binding protein (Mss4-MBP) was
affinity-purified on Amylose-Sepharose as described previously (13) and
then further purified to homogeneity by ion exchange chromatography on
Protein Pack 8HR.
Twenty-five micrograms of Mss4-MBP fusion protein was labeled with [35S]sulfur labeling reagent according to the manufacturer's instructions and separated from unincorporated label using Sephadex G25 resin. Purified His-tagged GTPases in Buffer A or urea (at a concentration of 2 mg/ml) were bound to 500 µl of Ni2+ resin and then extensively washed with Buffer A (50 mM Tris, pH 7.5, 100 mM NaCl, and 0.2% Tween 20). The amount of bound protein was verified by SDS-polyacrylamide gel electrophoresis. Ten µl of the beads were incubated with aliquots of radiolabeled Mss4-MBP in 200 µl of binding buffer (25 mM Hepes, 150 mM NaCl, 2.5 mM imidazole, 1 mg/ml bovine serum albumin, 0.2% Tween 20, pH 7.5) for 1 h at 25 °C. Beads were pelleted and resuspended in binding buffer containing 1% SDS, and radioactivity was measured by liquid scintillation counting. The specificity of binding was measured by competition with unlabeled Mss4-MBP (200 µg/ml).
Other TechniquesGel overlay assays with Mss4-MBP or
GTPS were performed as described previously (13) with the exception
that 35S-GTP
S was used instead of
[32P]GTP, and the concentration of dithiothreitol was
increased from 2 to 5 mM. [3H]GDP release
measurements were done using filter binding assays, which have been
previously described (13, 16). Co-precipitation assays between Mss4 and
Rab3a were performed as in Burton et al. (13), and the
presence of Rab3a was determined by Western blotting using anti-Rab3a
monoclonal antibodies (42.1) (kind gift of Dr. Reinhardt Jahn).
Mss4 has been
shown previously to bind and promote GDP-GTP exchange on only a subset
of Rab proteins belonging to the same branch of the Ras superfamily
(13). To address the specificity of this Mss4-Rab interaction, we
constructed several chimeras between a Rab that binds Mss4, Rab3a, and
one that does not, Rab5a (Fig. 1). Each DNA chimera was
subcloned into the bacterial expression vector pET15b, which adds six
histidines to the amino terminus of the expressed protein. The
recombinant proteins were produced in bacteria and tested for Mss4
binding by using the gel overlay technique described previously (13).
The recombinant Rab proteins were either purified by affinity
chromatography on Ni-NTA resin (soluble proteins left-hand panels (Fig.
2) or recovered from inclusion bodies and solubilized in
gel sample buffer (Fig. 2, right panels, Rab 3/5 IV and VII)
and electrophoresed on three identical SDS-polyacrylamide gel
electrophoresis gels. One gel was stained with Coomassie Brilliant Blue
to visualize the amount of protein loaded. Proteins from the other two
gels were transferred to nitrocellulose and overlaid with either
Mss4-MBP to determine Mss4 binding activity, or
35S-GTPS to determine GTP binding activity (Fig. 2).
We first tested the roles of the switch I region, the so called
effector domain (L2/2 according to the Ras crystal structure) and
switch II (L4/
2) domains of the Rab3a GTPase in Mss4 binding. We
reasoned that these switch domains of Rab3a may bind the GEF Mss4,
because they are proposed to undergo a dramatic conformational change
between the GDP- and GTP-bound state (17). In the Rab3/5 I and Rab3/5
II constructs, the switch I domain of Rab5 (Rab3/5 I) or both the
switch I and II domains of Rab5 (Rab3/5 II) were substituted with the
corresponding regions of Rab3a (Fig. 1). Both of these chimeric
proteins were unable to bind Mss4, suggesting that other Rab3a
sequences are necessary for the interaction (Fig. 2, middle
panel, data for Rab3/5 II not shown). We therefore constructed Rab3/5 chimeras in which the amino terminus of Rab5, including the
switch I (Rab3/5 III) or both the switch I and switch II domains (Rab3/5 IV) were replaced by the corresponding domains of Rab3a (Fig.
1). Both of these chimeric proteins were able to bind Mss4 (Fig. 2,
middle panels), although the construct lacking the switch II
domain was clearly less effective in binding (Fig. 2, middle right panel, compare Rab3/5 III with Rab3/5 IV; see also Fig. 3). The Rab3a switch I domain in combination with
sequences amino-terminal to this domain are crucial for Mss4
interaction because constructs Rab3/5 V and Rab3/5 VI, which lack the
switch I domain of Rab3a, do not bind Mss4 (Fig. 2).
In a separate study aimed at identifying Mss4-binding proteins by the
yeast two-hybrid system, a Rab3a clone lacking the first 29 aa from the
amino terminus was isolated (data not shown). We therefore attempted to
further define the Mss4-binding domain by removing these 29 aa from the
Rab3/5 III chimera. The resulting chimera, designated as Rab3/5 VII
(Fig. 1), which contains aa 30-73 of Rab3a and aa 72-215 of Rab5a,
was found to bind Mss4 albeit at considerably lower levels than what is
observed for Rab3/5 III, suggesting that aa 1-30 of Rab3a also
participate in Mss4 binding (Fig. 2, middle right panel).
The hypervariable domain located at the carboxyl terminus of Rab3a has
also been shown to be unnecessary for binding Mss4 in this yeast
two-hybrid screen (data not shown). Accordingly, two truncated forms of
Rab3a, Rab3aC1 and Rab3a
C2, bind to Mss4 at wild-type levels
(Fig. 2).
In order to quantitate the level of interaction between Mss4 and representative Rab3/5 chimeras, a solution binding assay was performed with 35S-labeled Mss4-MBP and His-tagged Rab3/5 III, Rab3/5 IV, Rab3/5 VII, Rab3a, and Rab5a bound to Ni2+ resin. Levels of Mss4 binding were ascertained by counting the radiolabeled Mss4 specifically associated with the Rab-bound Ni2+ resin. Rab3/5 IV bound Mss4 at levels approaching those of full-length Rab3a, whereas Rab3/5III and Rab3/5VII were much less efficient in interacting with Mss4 (Fig. 3). Therefore, the inclusion of the Rab3a switch II domain in Rab3/5 IV clearly increases Mss4 binding when compared with Rab3/5 III. These results are in full agreement with those found using the Mss4 overlay assay (Fig. 2). Taken together, our results implicate aa 1-73 of Rab3a, which includes the effector domain, as a critical region for Mss4 interaction, with residues 74-102 of Rab3a further enhancing Mss4 binding.
Note that Rab3/5 III, IV, and VII, which bind Mss4, do not bind significant levels of GTP by gel overlay (Fig. 2, bottom right panel) or in solution (data not shown). We have previously shown that the GEF Mss4 binds tightly to the GTPase in the absence of nucleotide, suggesting that it catalyzes GDP-GTP exchange by stabilizing the nucleotide-free form of the GTPase (13). Thus, a dissociation between Mss4-binding and GTP-binding properties of the GTPase is not unexpected.
Mss4 Point Mutation Disrupts Mss4 Binding Activity and GDP-releasing Activity on Rab3aTo complement the above results, we wished to address which domain of Mss4 is critical for binding and stimulating guanine nucleotide exchange on Rab proteins. Sequence comparisons of Mss4 and the related yeast protein Dss4 had shown that the region of highest similarity was a small stretch in the carboxyl terminus of the molecule, which includes conserved cysteines (10). Yu and Schreiber (18) have recently determined by nmr the structure for the human Mss4 protein. They found that the conserved cysteines (Cys23-X2-Cys26 and Cys94-X2-Cys97) enables Mss4 to bind Zn2+. Using site-directed mutagenesis, they showed that substituting a histidine for an aspartic acid at position 96 (Mss4 D96H) in this zinc-binding domain reduces Mss4-mediated GDP-release from Sec4 by 60%. However, the binding activity toward Sec4 was not investigated (18). We therefore produced Mss4 proteins harboring a point mutation at aa position 96 (Mss4 D96H) or 94 (Mss4 C94S) to determine how these mutations affected Mss4 binding activity and GDP-releasing activity toward Rab3a. Due to the instability of the Mss4 C94S protein, the biochemical properties of this protein could not be addressed. In contrast, Mss4 D96H was stable and was further investigated.
His-tagged wild-type Mss4 and Mss4 D96H proteins were linked to
Ni2+ resin and incubated with recombinant untagged Rab3a.
The beads were pelleted, washed, and assessed for the presence of Rab3a binding by Western blotting using Rab3a antibodies (Fig.
4, top panel). As shown previously, wild-type
Mss4 bound the nucleotide-free form of Rab3a, and this interaction was
disrupted in the presence of 1 mM GTP (Fig. 4, upper
panel, compare lanes 1 and 2; Ref. 13). In
contrast, the point mutant Mss4 D96H did not bind Rab3a, irrespective
of the presence of guanine nucleotide, although the amount of Mss4
protein present on the beads was similar for the Mss4 D96H and
wild-type Mss4 (Fig. 4, lanes 3 and 4,
upper left panel, lanes 1-4, upper right
panel). Quantitation of Rab3a protein bound relative to Mss4
present on the beads is depicted in Fig. 4 (bottom panel).
Mss4 D96H was 26-fold less efficient than wild-type Mss4 in binding to
Rab3a.
We next investigated the enzymatic activity of the Mss4 D96H mutant
protein toward Rab3a using the GDP release filter binding assay. We
found that an 8-fold molar excess of Mss4 D96H did not stimulate the
GDP-release rate from Rab3a, whereas wild-type Mss4 increased this rate
5-fold (Fig. 5). These results are consistent with the
lack of significant binding seen between Rab3a and the Mss4 D96H mutant
in the co-precipitation studies (see above). Taken together, these
results suggest that the aspartic acid at position 96 of Mss4 is
involved in binding the Rab GTPases.
In the present study, we have taken advantage of the selective
binding of Mss4 to a subset of Rabs to identify Rab domains important
for this interaction. Using chimeric Rab3/Rab5 proteins, we have
demonstrated that the binding requires the amino-terminal region of
Rab3a. The effector domain is crucially required for binding, while the
regions amino-terminal and carboxyl-terminal to this domain further
enhance binding. The mutation in the SEC4 gene
(sec4-8 allele), which is suppressed by a point mutation in
the DSS4 gene (11), maps to the 4 domain of the Sec4
protein, which is located toward the carboxyl-terminal domain of the
molecule (19). Thus, either the mutation in the Sec4-8 protein
(glycine to aspartic acid at aa position 147) has an indirect effect on the structure of the amino-terminal portion of the molecule, or Mss4,
which is a high copy suppressor, exerts its effects via an indirect
mechanism. For example, by stabilizing the nucleotide-free form of the
GTPase it may prolong the half-life of the Sec4-8 protein. The
carboxyl-terminal hypervariable domain of Rab3a is completely
dispensable for the interaction.
Strikingly, the region defined here as crucial for Mss4 binding is the
region that displays a higher degree of similarity among all of the
Rabs that had been previously shown to bind Mss4, and that is less
conserved in the other Rabs (13). As is shown in the alignment of Rab
protein sequences (Fig. 6), several residues in this
region are conserved in all of the Mss4-binding Rabs but not in other
Rabs or other members of the Ras superfamily (Fig. 6, shaded
residues). Rab13, which has been recently found to bind Mss4,2 also has these conserved amino acid
residues. These findings suggest that the property of binding Mss4 may
have been an important influence in the evolution of these proteins and
strongly supports the idea that Mss4 binding is critical for the
function of these Rabs.
It is of interest to compare our results with those of Macara et
al. (20, 21) concerning domains of Rab3a that are important for
interaction with Rab3a-GRF, a partially purified cytosolic exchange
factor whose primary structure is unknown. Using site-directed mutagenesis, they have found that mutations in the Rab3a switch I
domain decrease the affinity of Rab3a for Rab3a-GRF and cause the
protein to no longer be sensitive to Rab3a-GRF activity (20). In
addition, they have found that removal of the last 34 aa of Rab3a does
not affect Rab3a-GRF binding, both of these results are similar to
those reported here for Mss4 (20). In spite of these similarities,
several pieces of evidence suggest that Rab3a-GRF is not Mss4. First,
Mss4 and Rab3a-GRF differ in molecular mass. Whereas Mss4 is a 17-kDa
protein, the proposed molecular mass for Rab3a-GRF is about 295 kDa
(13, 21). Second, Mss4 binds to both lipid-modified
(geranylgeranylated) and -unmodified forms of Rab3a and promotes
GDP-GTP exchange on both proteins (13, 22). In contrast, Rab3a-GRF
prefers the lipid modified form (22, 23). Third, Rab GDI was reported
to be more effective in blocking the GDP-releasing activity of
Rab3a-GRF than of Mss4 when complexed with Rab3a (22). Finally, Mss4
has also been shown to stimulate GTPS release, whereas this activity
could not be detected for Rab3a-GRF (13, 23). Taken together, these studies suggest that Mss4 and Rab3a-GRF are distinct exchange factors,
although they interact with similar domains on Rab3a and may thereby
promote GDP release by an analogous mechanism. The Rab domains involved
in the binding to p619, a recently identified GEF for Arf1 and at least
some Rab proteins (12), remains to be identified.
Previous studies have investigated the domains of Ras that are
important for interaction with its GEFs, mSOS and Cdc25. The last 23 aa
of Ras have been shown to be dispensable for binding, in agreement with
the results reported here for the Rab3a-Mss4 interaction (24, 25).
However, a mutational analysis of Ras has shown that the switch II
domain (which consists of L4/2), but not the switch I domain, is
important for binding to its GEFs (26-28). Since Mss4 exhibits little
sequence homology with the GEFs for Ras, it is perhaps not surprising
that Mss4 and these GEFs interact differently with their respective
GTPases.
In this study we have further corroborated the key function of the carboxyl-terminal region of Mss4 in Rab-binding. This region, which contains a Zn2+-binding motif (18), is most highly conserved between Mss4 and its yeast homologue Dss4 (10). Furthermore, Dss4 was originally isolated by virtue of its ability to suppress the yeast sec4-8 strain when mutated in this region (D108G) (11). A single point mutation in this conserved domain of Mss4 (D96H) was previously shown to reduce GDP release from the nonphysiological substrate Sec4 by 60% (18). We show here that the same mutation not only abolishes Mss4-mediated GDP-release from Rab3a but also completely abolishes Mss4 binding to Rab3a.
In conclusion, we have defined domains on Rab3a and Mss4 that are
important for their reciprocal interactions. Our results suggest that
the Zn2+-binding face of Mss4 binds to the GTPase at the
region that surrounds its effector domain (1, L1,
1, L2,
2,
and L3). Conservation of this Mss4-binding interface appears to have
been a key constraint during the evolutionary diversification of a
subset of Rabs implicated in the secretory pathway (13).
We thank Dr. Marino Zerial for the generous gift of the Rab5a cDNA and Dr. Reinhardt Jahn for the Rab3a monoclonal antibodies.