STEM CELLS AND DEVELOPMENT 14:470–477 (2005)

© Mary Ann Liebert, Inc.

Issues in Development

Three-Dimensional Structure Prediction of the Interaction of

CD34 with the SH3 Domain of Crk-L

GURUDUTTA U. GANGENAHALLI,1,3,4 VIMAL K. SINGH,1,2,4 YOGESH K. VERMA,1,2

PALLAVI GUPTA,1 RAKESH K. SHARMA,1 RAMESH CHANDRA,2 SHWETA GULATI,1
and PRATIBHA M. LUTHRA2

ABSTRACT

The monomeric 115-kDa surface protein CD34, which is present on many stem cell populations, has
been useful to enumerate the quality and viability of cell suspensions for engraftment. Although
these studies assure the validity of CD34 as a stem cell marker, the functional role of this molecule
has not been defined. CD34 has been demonstrated to regulate adhesion, differentiation, and pro-
liferation of hematopoietic stem cells and other progenitors. The cytoplasmic domain of CD34 is
known to be essential for its function. However, it is not clear how this domain’s interactions with
other molecules support the functional activity of CD34. Here we show that the cytoplasmic tail of
CD34 is structurally similar to the carboxyl terminus of the gap junction protein Connexin 43 (Cx43).
Because the activity of CD34 is mediated through its interaction with an SH3 domain of an intra-
cellular protein, we attempted to define the SH3 binding region and amino acids involved in this in-
teraction. We identified Glu325 to Ser334 as potential SH3 binding sites. Our results suggest that
the interaction of the cytoplasmic tail of CD34 with the shallow proline-rich motif-binding groove
of Crk-L is essential for the function of CD34 in stem cell development.

INTRODUCTION that it is a highly glycosylated, type I integral protein. Other

studies have implicated CD34 in regulation of the adhesive

T HE PRESENCE OF THE SURFACE ANTIGEN CD34 on he-

matopoietic stem cells is presently used as a selective
marker for enumeration of these populations for transfu-
sion and engraftment for hematopoietic disorders (1). A
small subset of stem cells and progenitors that are capable
of regenerating hematopoietic cells from different onto-
logical sources, such as bone marrow, invariably express
this antigen (2–4). Ongoing efforts to unravel the structural
and biological significance of this molecule have shown
properties of hematopoietic progenitors (2,3,5) and in cy-
tokine-induced differentiation and secondary colony-form-
ing capability, as demonstrated in myeloid leukemia cell
line and normal CD34
progenitors in vitro, respectively
(6,7). These results, while not defining, have led to the as-
sumption that CD34 is essential for multidimensional func-
tions in hematopoietic milieu such as adhesion, differenti-
ation, and proliferation. However, the role of CD34 in stem
cell function remains to be elucidated.

1Stem-Cell Gene Therapy Research Group, Institute of Nuclear Medicine & Allied Sciences, Delhi-110054, India.
2Dr. B. R. Ambedkar Center for Bio-Medical Research, University of Delhi, Delhi-110007, India.
3Department of Hematology, Oncology and Stem Cell Therapeutics, School of Medicine, University of Pennsylvania, Phila-

delphia, PA 19104.
4These authors contributed equally to this work.

470
 CD34 MOLECULAR INTERACTION WITH CELLULAR PROTEINS
The predicted CD34 structure consists of a large ex-
tracellular domain (278 residues), followed by a stretch
of hydrophobic amino acids within the transmembrane
region (
22 residues) and a small cytoplasmic tail (73
residues) (3,5). The vast number of N/O-linked carbohy-
drates present on the extracellular domain are necessary
for stability, specific antigencity, and interaction of CD34
with counter-receptors or ligands. Although the cyto-
plasmic tail is devoid of any known enzymatic catalytic
domain, it possesses two consensus protein kinase C
(PKC) sites and one protein tyrosine kinase recognition
site at Thr356, Ser362, and Tyr318, respectively. The sto-
chiometric phosphorylation of the CD34 cytoplasmic tail
(CD34CT) by PKC at these sites is correlated with up-
regulated surface expression (3,5).
CD34 is recognized as a counter receptor of L-selectin
on vascular epithelial cells, (8) but not hematopoietic tis-
sues. Monoclonal antibodies to CD34 on hematopoietic
cells mimic its natural ligand and induce enhanced ad-
hesiveness (9). Deletion of the cytoplasmic domain of
CD34 abrogates its ability to up-modulate these adhesive
functions; thus, intracellular signaling regulation of the
adhesive functions of CD34 may play a key role in its
activity (7). CD34CT may rely on adapter proteins such
as Crk-L to participate in such a signaling cascade (10).
Complexes of Crk family proteins along with other mem-
bers such as Nck and homologs of Grb2/Sem5 are well
known in cellular signaling (11,12). The Crk-L gene,
identified by ten Hoeve et al., (13) shares
60% sequence
homology with the Crk-II, one of the two proteins gen-
erated by alternative splicing of human Crk proto-onco-
gene. Both proteins are related to v-Crk, the oncogenic
avian retrovirus CT10. Crk-L has been implicated in var-
ious biological activities, including signal transduction by
integrins, B cell and T cell receptors, and cytokines, such
as erythropoietin, interleukin-3, stem cell factor, and
thrombopoietin (14–16). Overexpression of Crk-L in
Ba/F3 cell lines and adhesion of hematopoietic cells to
fibronectin-coated surfaces (17) simultaneously demon-
strate a confirmatory functional link between Crk-L and
CD34.
The Crk-L protein consists of one SH2 and two SH3
domains, which may interact with large number of pro-
teins forming a multiprotein complex at the site of re-
ceptor activation. For example, integrin cross-linking
leads to induction of rapid tyrosine phosphorylation of
Crk-L and binding through its SH2 domain to other cel-
lular tyrosine phosphorylated proteins such as Cas, Cbl,
paxillin, etc. At the same time, the SH3 domain of Crk-
L constitutively binds to the Proline-X-X-Proline (P-X-
X-P) sequence present in various proteins, including c-
Abl, Dock 180, C3G, etc. (17,18). Proteins like Cbl and
Cas function as docking proteins, which interact with cel-
lular kinases such as 14-3-3 kinase and thus provide a
link between the nonkinase receptors and cellular kinases
471
(19,20). The Crk–Cas complex activates c-Jun kinase
through activation of p21rac, which is frequently associ-
ated with enhanced migration and adhesion (21).
SH3 domains are involved in a diverse range of pro-
cesses, and their structural investigations have been ini-
tiated at various levels from folding and thermodynam-
ics to protein ligand recognition and binding (22–25). It
has been established that SH3 domains bind to a polypro-
line helical ligand containing a consensus P-X-X-P se-
quence (24,25). Generally, the SH3 domain bears a rel-
atively flat, hydrophobic, ligand-binding surface, which
consists of three shallow pockets or a groove defined by
conserved aromatic residues. Most of the SH3 ligands
adopt an extended
-helical conformation termed as
polyproline type II (PP II) helix. The PPII helix has three
residues per turn, which means that it is roughly trian-
gular in cross section and the base of this triangle sits on
the surface of SH3 domain (24,25). SH3 domains are con-
sidered to bear three specific ligand binding pockets, and
two of them are occupied by two hydrophobic proline
(P) in dipeptide on two adjacent turns of the helix.
Whereas the third ‘specificity pocket’ in most cases in-
teracts with a basic residue in the ligand distal to the
PxP core. The helical conformation appears to play
an important role in binding of PPII helix to the SH3 sur-
faces (26,27). This fact is evident from the observation
of intramolecular binding of the Src SH3 domain to a
linker region between SH2 and catalytic domain of the
Src kinase, which adopts typical PPII helical conforma-
tion despite the absence of sequence similarity to known
Src SH3 binding sites (26). The helical conformation of
PPII ligand also provides freedom to bind in either “plus”
or “minus” orientations (24,25).
In view of the known CD34 interaction with Crk-L and
lack of knowledge about interacting regions in CD34 cy-
toplasmic tail, which are of biological significance, we
have explored details about SH3 domain of Crk-L inter-
action with the CD34 cytoplasmic tail. We have analyzed
interacting regions in both the proteins. Because the pri-
mary amino acid sequence of CD34 does not possess the
canonical P-X-X-P motif, we searched for non-P-X-X-P
ligands. Recently, sequences lacking the P-X-X-P motif
display the ability to bind to SH3 domains including se-
quences with the consensus P-X-X-D-Y derived from
e3bl/Abi-1 and RN-tre proteins (28), sequence with con-
sensus PX [VI][DN] RXXKP from the UBPY, Gab-1,
AMSH, and SLP-76 proteins (28, 29), sequence with the
consensus RKXXYXXY from the protein SKAP55 (30),
and sequence with the consensus WXXXFXXLE from
the protein Pex5p (31,32). These results provide essen-
tial support for the assumption that it is very likely that
interactions with the SH3 domain are essential for CD34
functions. The above-mentioned proteins have been dem-
onstrated to interact with a variety of proteins such as
Eps8 (recognizing consensus P-X-X-D-Y sequence),
 GANGENAHALLI ET AL.
Hbp, STAM, and Grb-2 (recognizing consensus PX
[VI][DN] RXXKP sequence), Fyb, Fyn, and Lck
(recognizing consensus RKXXYXXY sequence), and
Pex13p (recognizing consensus WXXXFXXLE se-
quence), respectively. However, the exact non-P-X-X-P
ligand-binding site may vary among different proteins.
To address the mode of CD34CT interaction with the
Crk-L SH3 domain, we examined the putative CD34CT
structure and explored distant homology to the gap junc-
tion protein Cx43 (33). The remote homology between
the two families (CD34 and connexins) was over 30%; a
similar secondary structural homology was detected in a
GRDB search (34). At the same time, we generated a
three-dimensional comparative model of CD34CT and in-
vestigated its ability to interact with the Crk-L SH3 do-
main. The molecular docking of this pair resulted in the
identification of a non-P-X-X-P SH3 binding region
flanking a region from Gly325 to Ser335 that is likely to
interact in the classical P-X-X-P ligand-binding groove.
The interaction between the two molecules might be sta-
bilized by a total of five hydrogen-bonding interacting
pair of CD34 residues: Glu325, Gly326, Tyr327, Ser334,
and Crk-L SH3 domain residues Asn146, Asp147, and
Tyr186 (see Fig. 4) and hydrophobic interacting pairs of
Thr333-Glu149, Pro331-Asp147, Glu325-Tyr186, and
Gly326-Phe143. In agreement with this observation, a
similar Crk-L SH3 domain residue has been reported to
be involved in the association of Crk-L with classical P-
X-X-P ligands of SH3 domain viz C3G and Sos (11,19).
Overall, the present study suggests the possible interac-
tion of CD34CT in the shallow proline-rich motif-bind-
ing groove of Crk-L and the sequence amino terminal to
the region described here might be supporting complex-
ation. These results are likely to facilitate the delineation
of the CD34–Crk-L interaction by providing an initial
guide for the development of specific and more sophis-
ticated exploring measures, which are a prerequisite for
elucidating the biological significance of the CD34–Crk-
L interaction in a hematopoietic intracellular milieu.
MATERIALS AND METHODS
Recognition of the CD34 putative fold
Alignment of CD34 protein sequence as a target (Gene
Bank-M81104) in the Gene Relate Sequence Database
(GRDB) (35) identified the putative homology fold for
CD34CT. The GRDB system contains the characteristic
profiles computed for many protein families collected
from Pfam cluster of ortholog groups, the Protein Data
Bank 7 (PDB 7), and other genomic sequences. GRDB
uses the MetaBASIC program to compare any protein,
like the CD34CT sequence, with about 100,000 protein
sequences in the database (35). MetaBASIC, unlike many
472
other methods used in fold recognition, does not require
any information about any native structure to draw the
comparison. This property of MetaBASIC helped us to
predict the structurally related protein structure on to
which the CD34CT model can be built, although it has
low sequence similarity in known databases.
Generation and validation of the CD34 three-
dimensional model
The three-dimensional model for the cytoplasmic tail of
CD34 was developed on SWISS-MODEL protein homol-
ogy modeling server (http://www.expasy.org/swissmod/
SWISS-MODEL) (36). The sequence alignment generated
by GRDB was used to predict the three-dimensional model
of CD34CT. The stereochemical quality of model was an-
alyzed by Procheck (37) and Whatcheck (38) and by pro-
grams on Structure Analysis and Validation Server
(http://nihserver.mbi.ucla.edu/SAVS). The quality of
model was further evaluated by calculating Root Mean
Square Deviation (RMSD) between the template and the
CD34CT model. The RMSD values were calculated by su-
perimposition of C
-atoms of CD34 model over the tem-
plate Cx43 (pdb1r5s; accession number, P8050), all-
against-all using the swisspdb viewer program (36).
Docking of CD34 cytoplasmic tail with Crk-L
SH3 domain
The CD34 and Crk-L SH3 domain interaction and their
molecular simulation was performed by using the pro-
tein–protein docking program Hex 4.2 (39). The three-di-
mensional model for the Crk-L protein structure (pdb1cka;
accession number, Q64010) was obtained from PDB. The
information about the SH3 ligand was used as an initial
guide. The C
atoms of the Tyr186 residue in Crk-L SH3
domain were chosen as ‘receptor’ origin and the CD34CT
as ‘ligand’. The expected ligand-binding site in Crk-L was
positioned close to the residues of CD34CT, and docking
was performed with receptor cutoff angle of 45°. Evalua-
tion of the local environment and residue contacts in the
docked complex was done by using Verify 3D program (40)
and Parameter Optimized Surface (POPS) server (41,42),
which provided assessment of the complex on residue level
and helped in locating misfolded regions and correct con-
formations. POPS program used the information about the
local secondary structure, solvent accessibility, and the frac-
tion of side chain area of the docked complex that is cov-
ered by polar atoms. The solvent-accessible surface area
(SASA), which is a useful index of protein–protein interac-
tion, was computed for the individual proteins in the free
and bound state. Beside SASA, composition of the interface
in terms of constituent residues and overall hydropathy and
surface compatibility, which are critical elements in any as-
sociative process, were also calculated.
 CD34 MOLECULAR INTERACTION WITH CELLULAR PROTEINS
FIG. 1. Topological representation of the CD34 structure and
related protein structures sharing structural similarity with the
cytoplasmic domain of wild-type CD34 protein. The circular
structure represents helices in all of the structures. Structures
compared include: (1) CD34, (2) gap junction protein (PDB1r5s),
(3) amine dehydrogenase (PDB1jmx), and (4) 60S ribosomal
protein (PDB1sliv0). The topological cartoons were generated
by TOPS program (43).
RESULTS
Recognition of the CD34 specific-fold and
validation of its three-dimensional structure
On using the sequence of CD34CT as a query, one re-
lated structure of the gap junction protein Cx43 from rat
(pdb1r5s; accession mumber, P8050) was found. This
protein also has a SH3 binding region and a PDZ do-
main-binding region separated by
100 residues (44).
Unlike most of the reported SH3 ligands, the commonly
formed left-handed type II polyproline helices were not
observed in Cx43 protein. Beside Cx43, weak structural
similarities was also detected between CD34 cytoplasmic
tail and rat kinesin (pdb2kinA; accession number, P56538),
FIG. 2. Alignment generated by GRDB. Predicted secondary structures share similarities with the secondary structures of the
known proteins (three-dimensional structures) in the database.
-Helices are encircled and -sheets are underlined.
473
arginine dehydrogenase from Pseudomonas putida (pdbI-
jmx; accession number, Q88EMO), and 60S ribosomal
protein (pdb1s1iV; accession number, P02606) from
yeast (Fig. 1) (43). The structural similarities were also
identified between these proteins by their comparison
through protein structural comparison program TOPS
(Topology Of Protein Structure) (43) (Fig. 1).
Multiple sequence alignment between consensus se-
quences of the CD34CT and the four families showed the
conservation of sequence patterns and secondary struc-
ture despite low amino acid identity (Fig. 2). On the ba-
sis of 30% sequence and structural similarity, Cx43 was
used to predict the three-dimensional structure of
CD34CT (Fig. 3). The model satisfied all stereochemical
restraints, and all of the marginal restraints were within
acceptable parameters (Table 1). The RMSD from the
ideal values for bond lengths, angles, and improper di-
hedral angles are given in Table 1. A Ramachandran plot
for the CD34CT model predicted 43.6% residues in the
CORE region and 74.4% residues in CORE and AL-
LOWED regions, whereas GENEROUS regions con-
tained 26.6 % residues and no residue was present in DIS-
ALLOWED regions.
Docking simulations of CD34 and Crk-L
molecular interaction
In the light of previous reports on CD34 interaction
with the Crk-L adapter protein and the fact that template
used in this modeling, Cx43 is a SH3 ligand. We ana-
lyzed their interaction and reported stable complex for-
mation between them. For docking CD34CT and Crk-L
three-dimensional structural coordinates were used.
There is evidence that the Crk-L SH3 domain contains a
ligand binding site between the RT loop, n-Src loop, and
residues Phe143, Asp147, and Tyr186 along with other
residues strongly interact with the ligand (44). Therefore,
we expected to see a docking solution in which these
residues were physically close to CD34CT. However, vi-
sual inspection of our docking solutions showed only one
orientation (rank 1 after clustering) with a close approach
of aforementioned residues. Hence, this orientation was
chosen as the only feasible solution.
Comparison of the SASA of the Crk-L–CD34 complex
with the combined area of two individual proteins indi-
cated a substantial overall reduction of 9% of SASA. Those
 GANGENAHALLI ET AL.
FIG. 3. Superimposition of CD34CT (Arg311-His367, light)
and template structure Connexin 43 (Ile 77-Ile132, dark). The
root mean square deviation between the structures is 0.10 Å.
The CD34 model starts with the Arg311 (out of 373 residues
of the wild-type CD34 protein) and ends at His367, whereas
template (PDB 1r5s) starts at Ile77 and ends at Ile132 (out of
381).
elements in SASAs that are buried on complex formation
are listed in Table 1. In the Crk-L–CD34 complex, a total
of 16 residues were identified whose SASAs reduced on
binding (Table 2). Of these, 9 residues (56.29%) were hy-
drophobic. Analysis of the hydrophobic surfaces of indi-
vidual protein residues revealed the presence of discrete
complementary hydrophobic patches.
The Crk-L–SH3-N domain interaction has been very
well demonstrated to polyproline (PP) helix II peptides
from C3G and Sos proteins (22). There were close fit in-
teractions between the hydrophobic side chain residues
presented by PPII helix and highly conserved hydropho-
bic residues (Phe141, Phe143, Pro183, and Tyr186) (22).
In our modeling studies, a total of five hydrogen bond
interactions were observed between CD34 residues
Gln325, Gly326, Tyr327, Ser334, and Crk SH3 N-do-
main residues Asn146, Asp147, and Tyr186 (Fig. 4). All
of the hydrogen bonds were in the range of 2–2.5Å, which
is stable enough to hold the protein in the SH3 ligand-
binding pocket. Among them, there were four hy-
drophobically interacting pairs of Thr333-Glu149,
Pro331-Asp147, Gln325-Tyr186, and Gly326-Phe143,
which fit in closer to each other within the docking dis-
tance range of 5Å.
DISCUSSION
The CD34 antigen present on the hematopoietic stem
cells/hematopoietic progenitor cells (HSCs/HPCs) is the
most widely used selective marker shown to have po-
tential for intrinsic signaling leading to an up-modulat-
ing effect on the adhesion of HPCs/HSCs. Donna et al.
(10) have described CD34 constitutive interaction with
the Crk-L adapter protein, which is a regulatory mole-
cule in transducing signals from proteins lacking intrin-
sic kinase activity (e.g., CD34) to other intracellular pro-
teins (e.g., Cas, Sos, etc.) necessary during adhesion
events (10,18,19). A membrane-proximal region of
CD34CT was shown to inhibit interaction of CD34CT and
474
Crk-L protein competitively in in vitro coimmunopre-
cipitation assays, which suggests the possible Crk-L in-
teracting region. However, validation with specific SH3
domain interacting residues and structural elements needs
to be determined. Here we have attempted to demonstrate
the possible residues and type of interaction between the
two proteins by applying current molecular simulation
techniques. The protein–protein docking of CD34CT and
the Crk-L SH3 domain implicated a region exactly fol-
lowing the sequence as described by Donna et al. flank-
ing Gln325-Ser334 in the membrane proximal region of
CD34CT.
Using standard methods, such as the Basic Local Align-
ment and Search Tool (BLAST), it was impossible to find
CD34 homologs with known structures. Our search of a
structural homolog for CD34CT resulted in identification
of a set of structurally similar proteins (Fig. 2). The gap
junction protein Cx43 (Fig. 2) shared 30% sequence ho-
mology with the CD34CT and was used as template for
the comparative modeling. Cx43 is a membrane-embed-
ded protein possessing a SH3 ligand region (P-X-X-P) and
a PDZ binding motif. There are several Ser/Thr phos-
phorylation sites in the protein. The protein is likely to
regulate opening and closing of gap junctions in a Ca2
-
and nucleotide exchange factor-independent manner. Sim-
ilarly, CD34CT possesses a consensus sequence for Tyr
and Ser/Thr phosphorylation. Primary sequence analysis
of CD34CT by the motif-predicting server scansite (45)
indicates a putative PDZ domain-binding region in the
distal part of the sequence (data not shown). CD34CT-me-
diated adhesion signaling is also independent of Ca2
and
ATP exchange. All of these are in favor of Cx43 selec-
tion as a template for the homology modeling.
A molecular docking study of CD34CT to the Crk-L
SH3 domain was performed to identify the putative SH3
binding region in the CD34CT. The SH3 domain consists
of two antiparallel -sheets packed at right angles to each
other (24). There are two SH3 variable loops, the RT and
n-Src loop flanking a SH3 ligand binding groove formed
TABLE 1. STEREOCHEMICAL PARAMETERS OF
THE CD34 CYTOPLASMIC TAIL MODEL
Stereochemical parameters CD34CT
Bond length (Å)a 0.014
Bond angles (°)a 2.120
Most favored – anglesb 17 (43.6%)
Additional allowed – anglesb 12 (30.8%)
Generously allowed – anglesb 10 (26.6%)
Disallowed – anglesb 0 (0.0%)
 Angle planarity st. devb 3.997
Bad contacts/100 residuesb 1.000
aRMSD from ideal values.
bPROCHECK values.
 CD34 MOLECULAR INTERACTION WITH CELLULAR PROTEINS

TABLE 2. PERCENTAGE EXPOSURE OF SASA OF THE SELECTED RESIDUES OF CRK-L AND CD34

Protein Residues

Crk-L Phe141; [57.5%], Phe143; [58%], Asp147; [26.3%], Glu149; [49.28%], Arg162; [65.8%], Asp163;
[47.2%], Trp169; [17.26%], Pro183; [34.09%], Pro185; [47.57%], Tyr186; [9.4%]
CD34 Gln25; [9.46%], Gly26; [8.8%], Ser 29; [8.8%], Gly30; [10.1%], Pro31; [56.5%], Thr33; [21.7%]

by hydrophobic residues (viz. Phe-141, Phe143, Pro183,
Tyr186, Trp169, and Pro185), which are well conserved
in the Src family. These hydrophobic residues interact
with the consensus P-X-X-P motif (33). The scrutiniza-
tion of final docking solution of CD34CT and Crk-L re-
vealed hydrogen bonding between the CD34CT residues
Gln325, Gly326, and Tyr327 with the hydrophobic
residue Tyr186 in the Src loop. There are two more H-
bonding interactions between Tyr327, Ser334, and Asn146
and Asp147 in the complex, respectively. The above-men-
tioned conserved residues in the SH3 domain have been
demonstrated to form a ridge on the surface on the SH3
groove that interacts with the ligand residues (23). The
spacing of side chain of these hydrophobic residues en-
ables them to fit close enough to residues of the ligands
(PPII helix) (13,20). In the CD34CT–Crk-L complex, a
close-fit interaction is visible among the residues Thr323-
Glu149, Pro331-Aso147, Glu325-Tyr186, and Gly326-

FIG. 4. (A) CD34–Crk-L interaction. CD34CR (right, light-shaded) and Crk-L (left, dark-shaded) were used as ligand and recep-
tor for the docking. The numbers of residues in both the molecules are shown in boxes. Interacting residues both in CD34 (Gln15,
Gly16, Pro21, Thr23) and Crk-L (Glu149, Asp147, Phe143, Tyr186) are highlighted in the space-filled version and the rest of the
molecules are in ribbon form. (B) Hydrogen bond interactions between CD34CT (Gln15, Gly16, Tyr17, Ser24) and the c-Crk SH3
N-domain (Tyr186, Asn146, Asp147). H bonds are shown as dotted lines and their distances are given in angstroms. The figure was
generated by the swiss pdb viewer. (The number indicates the residue number in the model; e.g., Gln15  Gln325 in the CD34 wild-
type protein. In the rest of the text, residues have been assigned their actual number in the CD34 wild-type protein only.)

475
 GANGENAHALLI ET AL.
Phe143. Similar interactions are reported in Crk-L and
C3G peptide binding where acidic residues Asp147,
Glu149, and Asp150 interact with the lysine residue of the
peptide. In contrast to the previously reported sequence
(10), an
-helical structure (Gly326-Gly322) could inter-
act in the classical SH3 active site, whereas the arginine-
rich region described by authors might interact in the other
regions of SH3 domain. This type of interaction is known
for the P-X-X-D-Y motif interaction with Eps 8 SH3 do-
main, which partially overlap specificity pocket and bind
in the distal loop region (28).
CONCLUSION
We first reported the existence of a putative three-di-
mensional structure for CD34CT. Despite the low se-
quence similarity, a possible three-dimensional fold of
CD34CT was generated based on its structural analogy
with remote homologs. Subsequently, this structure was
evaluated for its recognition capability as a counter-
receptor by adapter protein Crk-L SH3 domain. Molec-
ular simulation of CD34CT and Crk-L SH3 domain in-
teraction revealed the formation of a stable complex
involving CD34CT (Glu-325 to Ser-334) and Crk-L SH3
domain. To our surprise, CD34CT does not possess the
canonical P-X-X-P motif, which is likely necessary for
SH3 domain binding. However, this region attains an
-
helical conformation that like many of known SH3 do-
main ligands likely to facilitate their binding to relatively
flat hydrophobic groove on SH3 domains. Next, we iden-
tified residues likely to interact between two proteins and
report Glu325, Gly326, Tyr327, Thr333, Pro331, and Ser
334 as potential SH3 domain binding residues, which
seems to interact with certain residues in Crk-L known
to bind with C3G and Sos, etc. However, the exact mech-
anism of involvement of each residue in this interaction
needs to be elucidated by more sophisticated techniques
such as crystal structure solution or site-directed muta-
genesis. Nonetheless, the ability of the CD34CT model to
represent its almost accurate fold may useful in identifi-
cation of the other putative cellular proteins interacting
with its cytoplasmic tail.
ACKNOWLEDGMENTS
We thank Prof. Ramesh Chandra and Prof. Vani Bra-
hamchari of the Dr. B.R. Ambedkar Center for Bio-
medical Research, Delhi University, and the Director
of the Institute of Nuclear Medicine and Allied Sci-
ences (INMAS) in Delhi for support. Mr. Vimal Kishor
Singh particularly, thanks the Council for Scientific
and Industrial Research (CSIR), India, for research
fellowship.
476
REFERENCES
1. Burt KE. (1999). Clinical utility in maximizing CD34
cell
counts in stem cell graft. Stem Cells 17:272–276.
2. Krause DS, MJ Fackler, CL Civin and WS May. (1996).
CD34: structure, biology and clinical utility. Blood 87:1–13.
3. Civin CL, LC Strauss, C Brovall, MJ Fackler, JF Schwartz
and JH Shaper. (1984). Antigenic analysis of hematopoie-
sis III; A hematopoietic progenitor cell surface antigen de-
fined by a monoclonal antibody raised against KG-a cells.
J Immunol 133:157–165.
4. Bhatia M, JC Wang, U Kapp, D Bonnet and JE Dick.
(1997). Purification of primitive human, hematopoietic
cells capable of repopulating immunodefficient mice. Proc
Natl Acad Sci USA 94:5320–5325.
5. Lanza F, L Healy and DR Sutherland. (2001). Structural
and functional features of the D34 antigen: an update. J
Biol Regl Homeost Agents 15:1–13.
6. Gordan MY, SB Marley and R Davidson. (2000). Contact
mediated inhibition o f human hematopoietic progenitors
cell proliferation may be conferred by stem cell antigen
CD34. Hematol J 1:77–86.
7. Fackler M, DS Krause, OM Smith, CI Civin and WS May.
(1995). Full length but not truncated CD34 inhibits hemato-
poietic cell differentiation of MI cells. Blood 85:3040–3047.
8. Baumhueter S, SS Marks and H William. (1993). Binding
of L-selectin to the vascular sialomucin CD34. Science
262:436–438.
9. Lyn H, G May, K Gale, F Grosveld, M Greaves and T En-
ver. (1995). The stem cell antigen CD34 functions as a reg-
ulator of hematopoietic cell adhesion. Proc Natl Acad Sci
USA 92:12240–12244.
10. Donna MF, ML McVeigh, GT Hoehn, CI Civin and MJ
Fackler. (2001). The adapter protein Crk-L associates with
CD34. Blood 97:3768–3775.
11. Arai A, Y Nosaka, H Kohsaka, N Miyasaka and O Miura.
(1999). Crk-L activates integrin Mediated hematopoietic
cell adhesion through guanine nucleotide exchange factor
C3G. Blood 93:3713–3722.
12. Feller SM, R Ren, H Hanafusa and D Baltimore. (1994).
SH2 and SH3 domains as molecular adhesives: the inter-
actions of Crk and Abl. Trends Biochem. Sci. 19:453–459.
13. ten Hoeve J, C Morris and N Heisterkamp. (1993). Isola-
tion and chromosomal localization of Crk-L, a human Crk
like gene. Oncogene 8:2459–2474.
14. Barber DL, JM Mason, T Fukazawa, KA Reedquist, BJ
Druker, H Band and AD D’Andrea. (1997). Erythropoietin and
interleukin-3 activate tyrosine phosphorylation of CBL and as-
sociation with CRK adapter proteins. Blood 89:3166–3174.
15. Oda A, Y Miyakawa, BJ Druker, A Ishida, K Ozaki, H
Ohashi, M Wakui, M Handa, K Watanabe, S Okamoto and
Y Ikeda. (1996). Crk-L is constitutively phosphorylated in
platelets from chronic myelogenous leukemia patients and
inducible phosphorylated in normal platelets stimulated by
thrombopoietin. Blood 88:4304–4313.
16. Sattler M, R Salgia, G Shrikhande, S Verma, E Pisick, KVS
Prasad and DG James. (1997). Steel factor induces tyro-
sine phosphorylation of Crk-L and binding of Crk-L to a
complex containing c-kit phosphaatidylinositol, 3-kinase,
and p120 (CBL). J Biol Chem 272:10248–10253.
 CD34 MOLECULAR INTERACTION WITH CELLULAR PROTEINS
17. Uemura N and JD Griffin. (1999). The adapter protein Crk-
L links Cbl to C3G after integrin ligation and enhances cell
migration. J. Biol. Chem 274:37525–37532.
18. Nosaka Y, A Arai and N Miyasaka. (1994). Crk-L medi-
ates Ras-dependent activation of the Raf/ERK through gua-
nine nucleotide exchange factor C3G in hematopoietic cells
stimulated with erythropoietin or interleukin-3. J Biol
Chem 272:30154–39162.
19. Robertson H, WY Langdon, CB Thien and DDL Bowtell.
(1997). A c-Cbl yeast two hybrid screen reveals interac-
tions with 14-3-3 isoforms and cytoskeletal components.
Biochem Biophys Res Commun 240:46–50.
20. Liu Y, Y-C Liu, N Meller, L Giampa, C Elly, M Doyle and
A Altman. (1999). Protein kinase C activation inhibits ty-
rosine phosphorylation of Cbl and its recruitment of Src
homology 2 domain-containing proteins. J Immunol 162:
7095–7101.
21. Dolfi F, M Garcia-Guzman, M Ojaniemi, H Nakamura, M
Matsuda and K Vuori. (1998). The adaptor protein Crk con-
nects multiple cellular stimuli to the JNK signaling path-
way. Proc Natl Acad Sci USA 95:15394–15399.
22. Wu X, B Kudsen, SM Feller, J Zheng, A Sali, D Cowburn, H
Hanafusa and J Kuriyan. (1995). Structural basis for the spe-
cific interaction of Lysine containing praline-rich peptides with
the N-terminal SH3 domain of c-Crk. Structure 3:215–226.
23. Musacchio A, M Saraste and M Wilmanns. (1994). High-
resolution crystal structure of tyrosine kinase SH3 domain
complexed with proline rich peptide. Nature Struct Biol
1:546–551.
24. Kay BK, MP Wiliomson and M Sudol. (2000). The im-
portance of being proline: the interaction of proline rich
motifs in signaling proteins with their cognate domains.
FASEB J 14:231–241.
25. Mayer BJ. (2001). SH3 domains: complexity in modera-
tion. J Cell Sci 114:1253–1263.
26. Williams JC, A Weijland, S Gonfloni, A Thompson, SA
Courtneidge, G Superti-Furga and RK Wierenga. (1997).
The 2.5 Å crystal structure of the inactivated form of
chicken Src, A dynamic molecule with muliple regulatory
interactions. J Mol Biol 274:757–775.
27. Moniovi AM, PR Romano, S Panni, M Mendoza, WT
Wong, A Musacchio, G Cesareni and P Di Fiore. (1999).
A novel peptide SH3 interaction. EMBO J 18:5300–5309.
28. Kato M, K Miyazawa and N Kitamura. (2000). A deubiq-
uitinating enzyme UBPY interacts with the Src homology3
domain of Hrs binding protein via a Novel binding motif
PXP[V/I][D/N]RXXKP. J Biol Chem 275:37481–37487.
29. Lewitzsky M, C Kardinal, NH Gehring, EK Schmidt, B
Konkol, M Eulitz, W Birchmeier, U Schaeper and SM
Feller. (2001). The C-terminal SH3 domain of the adapter
protein Grb2 binds with high affinity to sequence in Gab1
and SLP-76 which lack the SH3 typical PxxP core motif.
Oncogene 20:1052–1062.
30. Kang H, C Freund, JS Duke-Kohan, A Musacchio, G Wag-
ner and CE Rudd. (2000). SH3 domain recognition of a
proline independent tyrosine based RKxxxYxY motif in
immune cell adapter SKAP55. EMBO J 19:2889–2899.
31. Barnett P, G Bottger, TJA Klein, HF Tabak and B Distel.
(2000). The Peroxisomal membrane Protein Pex13p shows
a novel mode of SH3 interaction. EMBO J 19:6382–6391.
477
32. Urquhart AJ, D Kennady and SJ Gould. (2000). Interaction
of Pex5p,the type I peroxisome target receptor, with the
Peroxisomal membrane proteins Pex14p and Pex 13P. J
Biol Chem 275:4127–4136.
33. Sorgen PL, HS Duffy, P Sahoo, W Coombs, M Delmar and
DC Spray. (2004). Structural changes in the carboxyl ter-
minus of the gap junction protein connexin 43 indicates
signaling between binding domain for c-Src and zonila oc-
cludens-1. J Biol Chem 279:54695–54701.
34. Ginalski K, M Grotthuss, NV Grishin and L Rychlewski.
(2004). Detecting distant homology with MetaBASIC. Nu-
cleic Acids Res 32 (Web server issue):W576–W581.
35. von Grothuss M, PJ Wrrwicz, K Ginalski and L Rych-
lewski. (2003). Application of the 3-D jury, GRDB and
Veryfy 3D in fold recognition. Proteins 53:18–423.
36. Guesx N, A Diemand and MC Peitsch. (1999). Protein
modeling for all. Trends Biochem 24:64–67.
37. Laskowaski RA, MW MacArthur, DS Moss and JM Thor-
ton. (1993). PROCHECK: a program to check the stereo-
chemical quality of protein structure. J Appl Cryst
26:281–293.
38. Hooft RWW, G Vriend, C Sender, and EE Abola. (1996).
Errors in protein structures. Nature 381:272–272.
39. Ritchie DW. (2003). Evaluation of protein docking pre-
dictions using Hex 3.1 in CAPRI rounds 1 and 2. PRO-
TEINS: Struct Funct Genet 52:98–106. (Accessible at
http://www.csd.abdn.ac.uk/hex/.
40. Luthy R, JU Bowie, and D Eisenberg. (1992). Assessment
of protein models with three-dimensional profiles. Nature
356:83–85.
41. Fraternali F and L Cavallo. (2002). Parameter optimized sur-
faces (POPS): Analysis of Key interactions and conformational
changes in the ribosome. Nucleic Acids Res 30:2950–2960.
42. Cavallo L, J Kleinjung, and F Fraternali. (2003). POPS: a
fast algorithm for solvent accessible surface areas at atomic
and residue level. Nucleic Acids Res 31:3364–3366.
43. Michalopoulos I, GM Torrance, DR Gilbert and DR West-
head. (2004). TOPS: an enhanced database of protein struc-
tural topography. Nucleic Acids Res 32:D251–254, data-
base issue.
44. Smit L, G van der Horst and J Borst. (1996). Sos, Vav and
C3G participate in B-cell receptor-induced signaling path-
way and differentially associates with shc-Grb2, Crk and
Crk-L adapters. J Biol Chem 271:8564–8569.
45. Yaffe MB, GG Leparc, J Lai, T Obata, S Volinia and LC
Cantley. (2001). A motif-based profile scanning approach
for genome wide prediction of signaling pathways. Nature
Biotechnol 19:348–353.
Address reprint requests to:
Dr. G.U. Gangenahalli
Chief, Stem-Cell Gene Therapy Research Group
Institute of Nuclear Medicine & Allied Sciences
Lucknow Road
Delhi-110054, India
E-mail: gugdutta@rediffmail.com
Received November 1, 2004; accepted August 15, 2005.