Emerging Roles for Crk in Human Cancer

Introduction

Named for their ability to increase cellular tyrosine phosphorylation and transform primary chicken embryo fibroblasts (CEFs), CT10 regulator of kinase (Crk) is the prototypical member of a class of proteins called adaptors that contain Src homology-2 (SH2) and Src homology-3 (SH3) domains but lack intrinsic enzymatic activity. First discovered by Mayer and Hanafusa as the transforming gene product of an avian retrovirus, ¹ the cellular homologs CrkI and CrkII ^2,3 and the related Crk-like (CrkL) ⁴ are expressed in all tissues and identified across eukaryotic organisms. Over the past 2 decades, these ubiquitously expressed proteins have been implicated in complex and diverse physiological processes, ranging from regulation of the actin cytoskeleton in cell motility, phagocytic entry of apoptotic cells and pathogens into host cells, as well as cell cycle, apoptosis, and metabolism. ^5,6 A general theme has also emerged that Crk proteins are dysregulated in several human malignancies, which will be the subject of this review.

There are several distinct members of the Crk family in addition to the retrovirus-encoded v-Crk. CrkI and CrkII were originally described as splice variants, ² where CrkI resembled a structure analogous to v-Crk while CrkII possessed an extended C-terminus containing a proline-rich linker and an atypical SH3 domain (SH3C), which does not bind to classic proline-rich motifs ^7,8 (Fig. 1). More recently, a third splice variant of crk, called CrkIII, has been described that is predicted to encode a protein with a truncated SH3C, ⁹ but the biological significance of this remains unknown. By contrast, CrkL has not been reported to be alternatively spliced. Despite the fact that CrkII and CrkL have a high degree of similarity, it is still not well understood how these isoforms vary in ligand affinities and specificity or how the 3-dimensional structures of CrkII and CrkL differ to engage key signaling partners. There are indeed expected to be differences in the signaling characteristics and/or tissue-specific expression of Crk and CrkL given the fact that the respective knockout mice have distinct phenotypes and that both proteins are required for embryonic development. ^10,11

OPEN IN VIEWER

Figure 1.

Schematic of the Crk family of adaptor proteins. Common protein interacting domains of the Crk proteins are indicated in comparison with Src. Gag refers to the group-specific antigen encoded by the CT10 avian retrovirus. Y527 in Src, Y221 in CrkII, and Y207 in CrkL are regulatory tyrosine motifs that bind intramolecularly to the SH2 domain.

Adaptors and Assembly of Signaling Complexes

The primary mechanism by which Crk and CrkL function is through the organized assembly of protein-protein complexes in a phosphotyrosine-dependent manner. ⁵ Shortly after the original publication describing Crk as an oncogene with sequence similarity to the N-terminal noncatalytic region of Src family kinases, Matsuda et al. reported for the first time that SH2 domains possess general phosphotyrosine binding activity ¹² and that both the SH2 and SH3 domains were required for Crk signaling. ¹³ It was later realized that while the Crk SH2 domain recognized targets in a phosphotyrosine-dependent manner, speci­ficity and high-affinity interactions were determined by amino acids carboxyl-terminal to the pTyr with the consensus pTyr-X-X-Pro. ^14-16 A few years later, after several SH3N binding proteins were characterized, Knudson et al. used organized proline-oriented peptides to identify SH3 binding targets, and the consensus sequence for the Crk SH3N domain was determined to be Pro-x-x-Pro-x (Lys,Arg). ¹⁷ Subsequently, phage display screening technologies extended this paradigm of SH3 domain speci­ficity, and several variations on the Pro-x-x-Pro/SH3 domain interactions emerged. ^18,19 All Crk and CrkL binding partners, while diverse in their molecular function, utilize these common building blueprints to achieve a great degree of diversity in cell signaling. In contrast, the C-terminal SH3 domains of Crk and CrkL are atypical in that they do not bind polyproline type II (PPII)–containing motifs. ⁷ Indeed, the Crk SH3C domains are thought to confer allosteric regulation either on Crk or on other proteins. ^20,21 This has raised the possibility that the atypical binding sites on Crk and CrkL SH3C may be involved in either lower affinity interactions or interactions that can be subjected to novel means of regulation.

Insight into the regulation of interactions between Crk and its ligands first emerged from biochemical studies ^22,23 and then was confirmed by NMR ²⁴ and X-ray crystallography. ²⁰ These studies demonstrated that the biological activity of CrkII was controlled by tyrosine phosphorylation on Y221 by Abl (CrkL on Y207) that resulted in intramolecular autoinhibition. Such autoinhibition displaces the SH2 domain from binding phosphotyrosine motifs in trans. ²⁰ This autoinhibition was notable for 2 important reasons: first, it clearly identified a mechanism by which v-Crk acquired transforming potential by deletion of the C-terminal–negative regulatory element; and second, it provided an analogous mechanism for the oncogenic activation of v-Crk as to the oncogenic activation of v-Src (deletion of Y527). ^25,26 This pattern of activation of Crk and Src provided paradigms for the oncogenic activation of proto-oncogenes, a legacy established in the Hanafusa laboratory.

Crk in Human Cancers

Following the initial characterization of v-Crk as a transforming oncogene from the avian CT10 retrovirus, it became immediately clear that both v-Crk and CrkI possessed dominant oncogenic properties in chicken embryo fibroblasts, mouse NIH 3T3 cells, and when cells were injected subcutaneously into nude mice. Overexpression of CrkII ² or CrkL ^27,28 also transformed mammalian cells, although “weaker” in activity compared to CrkI mainly due to the negative regulatory tyrosine phosphorylation in their carboxyl-termini. However, forced CrkII expression recapitulates several features of malignant transformation, including alterations in focal adhesions, induction of Rac1-dependent membrane ruffles and lamellipodia, and altered adhesion. The original observations that a scaffolding protein such as v-Crk could transform cells implied that Crk was not simply a passive conduit for signaling but could amplify signaling output. In the last decade, the role of CrkII and CrkI in human cancers has steadfastly gained credibility. Both these proteins are frequently associated with aggressive and invasive human cancer phenotypes. Below, we review the evidence that Crk plays an important role in human cancer (Table 1).

OPEN IN VIEWER

Table 1.

Summary of the Analyses of Clinical Specimens for Crk and/or miR-126 Expression in Human Cancers

Cancer Type: Lung Cancer (Miller et al. 30 )
Classification	Normal	Stage I	Stage III
Total Crk mRNA (sample size)	CNA	+ (67)	++ (19)
Crk protein levels
CrkI	+	++	+++
CrkII	+	+	+
pCrkII	+	++	++
Sample size	10	57	19
Classification	Well Differentiated	Moderately Differentiated	Poorly Differentiated
Total Crk mRNA	+	+	++
Crk protein levels
CrkI	+	+	++
CrkII	+	++	+++
pCrkII	+	++	+++
Sample size	22	47	23
Cancer Type: Breast Cancer (Rodrigues et al. 32 )
Classification	Normal	Stage I	Stage III
No. of samples	— a	8	12
Crk protein level elevated	— a	63%	58%
Cancer Type: Glioblastoma (Wang et al. 41 )
Classification	Normal Brain Tissue	Benign (Grade I, II)	Malignant (Grade III, IV)
Crk mRNA level
CrkI	+	+	++ (8 cases)
CrkII	+	+	++ (3 cases)
Sample size	1	4	16
Cancer Type: Gastric Cancer (Feng et al. 58 )
Classification	Local Invasion: T1, T2	Local Invasion: T3, T4
miR-126 expression (n = 60)	Low (7 cases), high (15 cases)	Low (26 cases), high (12 cases)
Classification	Lymph Node Metastasis: No	Lymph Node Metastasis: Yes
miR-126 expression (n = 60)	Low (6 cases), high (13 cases)	Low (27 cases), high (14 cases)
Classification	TNM Stage I, II	TNM Stage III, IV
miR-126 expression (n = 60)	Low (4 cases), high (18 cases)	Low (29 cases), high (9 cases)

Note: Crk mRNA levels were analyzed either by Affymetrix oligonucleotide arrays (in lung cancer) or semiquantitative RT-PCR (in glioblastoma), and Crk protein levels were analyzed either by 2D-PAGE and Western blotting (in lung cancer) or immunohistochemistry (lung and breast cancer). miR-126 expression was analyzed by qRT-PCR. CNA = comparison not available; TNM = tumor node metastasis.

a

Staining of adjacent normal epithelia was used for comparison.

Sarcomas

Synovial sarcoma is a rare malignant tumor of soft tissues that typically surround joints, tendons, muscles, blood vessels, and other fibrous tissues. It is particularly aggressive and metastatic and quite early becomes lymph node positive. In several sarcomas, including synovial sarcoma, amplified HGF/c-MET signaling has been implicated in tumor progression including EMT, tumor growth, and metastasis to the brain and lung. ^42,43 Crk plays a central role in c-Met signaling ³⁴ by binding to the tyrosine-phosphorylated scaffolding protein Gab1. ⁴⁴ Binding of Crk to Gab1 sustains phosphorylation of Gab1 on Y307 that in turn induces activation of the Rac1 GTPase to promote cell migration (Fig. 2). Functionally, siRNA-mediated knockdown of Crk attenuated HGF-induced motility in human synovial sarcoma cell lines and also suppressed tumor formation in mouse xenograft studies. Subsequent studies further revealed that Crk promotes G1-S progression in synovial sarcoma cells mainly through the suppression of the CDK inhibitor p16 via Src-dependent activation of p38 MAPK. ⁴⁵

OPEN IN VIEWER

Figure 2.

Feedback regulation of Crk and signaling pathways activated in malignant cells. Crk is engaged in multiple signaling pathways that drive the malignant behavior of cells. Tyrosine-phosphorylated p130cas (via integrins) and Gab1 (via Met receptor) act as molecular switches to assemble complexes of Crk with DOCK180, C3G, or Sos to stimulate indicated phenotypic outcomes. Elevation in Crk expression can also feedback and activate FAK and Src to further increase phosphorylation of p130cas in an amplification loop. Binding of Crk to tyrosine-phosphorylated Gab1 sustains phosphorylation on Y307 to possibly amplify downstream signaling. Activation of Src inhibits miR-126 defining another amplification loop. Finally, activation of Abl from upstream pathways including the engagement of Eph receptors can disassemble signaling complexes by inducing Crk phosphorylation on Y221. Similar pathways may operate for CrkL.

miR-126 Regulates Crk Expression

Recently, a mechanism that may explain how Crk becomes up-regulated in cancer cells has emerged in the form of identification of a micro-RNA (miR-126) that targets the Crk mRNA. Indeed, expression of miR-126 decreased Crk expression at the protein level and functionally resulted in the attenuation of malignant phenotypes of lung cancer cells by inhibiting invasion and migration. ⁵⁶ Further studies showed that miR-126 was strongly suppressed in Src-transformed cells, offering a mechanism by which there is crosstalk between Src and Crk to influence tumor growth and invasion. ⁵⁷ miR-126 was also identified in a screen of human gastric cancer patients to be significantly down modulated, ⁵⁸ suggesting that miR-126 functions as a tumor suppressor in human gastric cancer. Finally, in the same study, miR-126 expression reduced Crk protein levels and inhibited cell invasion, migration, and metastasis in a mouse xenograft model.

Is Crk a Biomarker for Malignant Cancer Cells?

Do the above arguments and conclusions drawn from expression and signal transduction studies in a variety of human cancers imply that Crk and/or CrkL may be general biomarkers that predict aggressive tumor behavior? Early studies by Nishihara et al. evaluating human cancers showed that CrkII was significantly elevated in human lung and colon cancers, and typically expression correlated with lesion grade. ⁵⁹ Similar findings are emerging with the assistance of iTRAQ-based mass spectrometry and microarray analysis, where Crk has been identified as a metastasis-associated biomarker in lymphatic and breast cancers. ⁶⁰ More recently, Crk and CrkL were identified as “essential cancer causing genes” using genome-wide shRNA library screening platforms combined with computational approaches from 12 cancer cell lines representing diverse cancer types. ⁶¹ These data are in agreement with the general idea that Crk and/or associated signaling components may be potential prognosis markers and therapeutic targets.

Mechanisms and Interventions

Although a general pattern of Crk overexpression correlating with the malignant features of cancer cells has already emerged, extending the expression and experimental analyses of Crk to a much larger set of cancer cell lines and cancer tissues will permit a more systematic assessment of the extent to which Crk is overexpressed and its role in human cancers. Efforts to categorize Crk signaling in relation to specific cancer phenotypes will eventually lead to an evaluation of Crk and/or associated signaling components as therapeutic targets. Given the centrality of Crk in the aforementioned invasive tumor phenotypes, it is conceivable that introduction of miR-126 or shRNA to Crk into tumor cells might have therapeutic value in attenuating metastasis of invasive tumors. Proof of concept experiments in mouse models will provide an estimate of the likelihood of success of these strategies.

Further studies are also required to dissect out the exact mechanisms by which Crk becomes up-regulated in human cancer and how it drives aggressive phenotypes. There is a large body of evidence on investigating the latter, as complexes of either p130Cas/Crk/DOCK180 or Gab1/Crk/DOCK180 lead directly to Rac1 activation, altered actin cytoskeletal assemblages, and motility via interactions with WAVE and Arp2/3 complexes ⁵ (Fig. 2). Crk can also bind to C3G via its SH3N domain, and the resultant p130Cas/Crk/C3G complexes can activate Rap1-inducing inside to outside signaling, leading to increased integrin activation, p130Cas phosphorylation, and enhancement of cell motility. ⁶² It is noteworthy that the above pathways can be amplified by positive feedback loops established by Crk up-regulation, such as what is detectable in various human cancers. Forced expression of CrkII results in enhanced activation of FAK that further increases p130Cas and paxillin phosphorylation ¹⁴ or enhanced Gab1 phosphorylation via Src, ⁴⁷ thereby enhancing cell motility. Such amplification loops may not only enhance motility but can activate Src, ⁴⁵ which in turn may down-regulate miR-126 and increase Crk levels ⁵⁷ in a second amplification loop.

In a recent study to elucidate which signaling pathways of Crk mediate cell growth, Mayer et al. employed a systematic approach in CrkI-transformed NIH3T3 cells by individually knocking down SH3N binding proteins, including DOCK180, C3G, SOS1, and Abl/Arg. ⁶³ RNAi-mediated knockdown of C3G or SOS1 suppressed anchorage-independent growth and tumor formation in nude mice, while knockdown of DOCK180 had little effect on tumor growth. These data suggest that motility (DOCK180-dependent) and tumorigenicity (SOS1- and C3G-dependent) are separable in the NIH3T3 model, and such systematic knockdowns in other cancer cells would add important insight into which pathways contribute to specific phenotypes.

Role of Abl in Crk Signaling

While Abl is the sole tyrosine kinase that interacts with Crk SH3N, its role in Crk signaling during tumor progression remains surprisingly elusive. On one hand, in the Mayer study, ⁶³ knockdown of Abl in CrkI-transformed NIH 3T3 cells resulted in enhanced anchorage-independent growth and tumorigenic potential. Such observations are consistent with studies by the Pasquale laboratory, ³⁷ showing that Abl mediates the antitumorigenic effects of EphrinB2 in breast cancer by negatively regulating CrkII-mediated signaling. On the other hand, equally compelling evidence also suggests that Abl is activated downstream of deregulated EGFR and Her2/ErbB2 in aggressive breast cancer lines and enhances the invasiveness of those cell lines. ^64,65 However, reconciliation of this role of Abl with that of mediating the antitumorigenic effects of EphrinB2 by phosphorylation of CrkII at Y221 is not obvious as both tumor-suppres­sive EphrinB2/EphB4 signaling ³⁷ and mitogenic/invasive EGF/EGFR signaling have been shown to induce phosphorylation of CrkII at Y221. ⁶⁶ Conceivably, Crk may play a yet unrealized role in either or both signaling pathways that impinges on the balance between EphrinB4 and EGF signaling in breast cancer.

Special Remarks

The authors and editors dedicate this review to Hidesaburo Hanafusa, head of the Laboratory of Viral Oncology at Rockefeller University from 1973 to 2000. While initially dedicated to the biology of Src in the 1970s and 1980s, much of the effort in the Hanafusa laboratory shifted away from Src and toward Crk after the discovery of v-Crk in 1988. As fate would permit, conceptual advances on Src and other retroviral oncogenes studied by Hanafusa paved, in a large way, the road to unraveling the function of Crk in cell transformation. Besides obvious similarities in the composition of modular SH2 and SH3 domains in Crk and Src, both genes were similarly “hijacked” by RNA tumor viruses, and both exhibited C-terminal truncations that resulted in the loss of SH2-pTyr–mediated negative regulatory elements. These mechanisms remain today as textbook explanations for oncogene activation and the oncogene theory of human cancer.

The emerging importance of Crk in a vast variety of human cancers is reminiscent of the history of Src and the revelation that Src is important in human cancer. In this respect, Src was not always appreciated as a relevant player in human cancer. In the early 1990s, it was still not convincing whether Src had a significant role in human cancer since it was rarely mutated and not at all on Tyr527 in tumor samples. Of course, the evidence for a role of Src in human cancers is now quite compelling, and Src kinase inhibitors are entering the mainstream of clinical cancer therapy. ⁶⁷ The story of Crk in cancer therapeutics remains to be completed, although it is now clear to play a central role in malignant transformation.