Structure, function, and recombinant production of EGFL7

1.1 EGFL7 gene and protein structure

EGFL7 was first cloned as vascular endothelial station (VE statin) by Soncin and colleagues in 2003 from the 3′ end of vezf1 (Soncin et al. 2003). It is a rather small protein composed of multiple domains involved in cell-extracellular matrix (ECM) interactions (Nichol and Stuhlmann 2012; Schmidt et al. 2007a). The gene egfl7, encoding for the EGFL7 protein, is located on chromosomes 2 (2B) in mice and 9 (9q34.3-qter) in humans, with both the mouse and human egfl7 gene containing 11 exons. Human egfl7 contains two alternative exons, 1a and 1b, with transcription of the EGFL7 protein able to occur from different promoters, giving rise to alternative isoforms. However, these isoforms contain the same open reading frame, beginning at exon 3 and encompassing the rest of the gene (Fish et al. 2008). In addition, binding sites for the transcription factors ETS-related gene and GATA-2 (Bras et al. 2010) are present at the egfl7 gene locus. Of further note, egfl7 harbors the microRNA-126 within intron 7 in both humans and mice. MicroRNA-126 is a small non-coding RNA expressed, among others, in endothelial cells and implicated in modulating gene expression associated with angiogenesis and vascular integrity (Ebrahimi et al. 2014; Meister and Schmidt 2010). The expression of microRNA-126 is regulated by the transcription factor Krüppel-like factor 2 (Harris et al. 2010). The open reading frame of the human egfl7 encodes for a protein of 273 amino acids in length and a molecular weight of 29.6 kDa. The mouse homolog is approximately the same size as the human EGFL7 protein and is 275 amino acids in length with a molecular weight of 29.8 kDa. The proteins display a molecular weight of about 41 kDa upon post-translational modifications.

With regard to protein structure, EGFL7 is a modular assembled protein composed of multiple domains. From N-terminus to C-terminus, EGFL7 consists of a signal peptide, an Emilin-like domain, two EGF-like domain repeats, and a conserved coiled-coil domain at the C-terminus. The presence of the signal peptide indicates that the protein is targeted for secretion into the extracellular space (Soncin et al. 2003). The cysteine-rich Emilin-like domain is found in secreted proteins of the Emilin family and is involved in cell to cell interactions and ECM binding (Doliana et al. 2000). EGF repeats are common in extracellular proteins and are involved in numerous adhesion and ligand/receptors interactions. For example the more N-terminal EGF-like domain is implicated in non-canonical Notch receptor signaling (Schmidt et al. 2009), while an associated arginylglycylaspartic acid (RGD) motif facilitates EGFL7’s ability to bind integrins, in particular α_vβ₃ and α₅β₁ (Dudvarski Stanković et al. 2018; Nikolić et al. 2013). The more C-terminal EGF-like domain binds Ca²⁺ and is proposed to induce a conformational change in the protein (Fitch et al. 2004; Lindsell et al. 1995). The coiled-coil domain at the C-terminus has been suggested to act as a “retention signal”, inhibiting EGFL7 secretion from HEK-239 cells, with the expression of constructs missing the coiled-coil domain facilitated EGFL7 detection in the extracellular conditioned media (Picuric 2007). This finding indicates that post-translational modifications (e.g. proteolytic processing) may be required for secretion in the protein’s native environment. EGFL7 also has both a positively charged N-terminus and a negatively charged C-terminus, with this polarity allowing for oligomer formation in the ECM following secretion. Various ECM proteins including collagen type I and fibronectin facilitate this process (Dudvarski Stankovic et al. 2018; Nikolić et al. 2013; Schmidt et al. 2007b). Finally, the EGFL7 protein undergoes significant post-translational modification prior to secretion from expressing cells. These include post-translational modifications at the Emilin-like domain, N-terminus, and C-Terminus (see Table 1), however information regarding the function of these modifications is limited. No post-translational modifications have been reported for either of the EGF-like domains in EGFL7 in humans or mice. We note, however, that functional post-translational modifications do occur in other proteins containing EGF-like domains, such as in Notch (Wouters et al. 2005) and extracellular matrix proteins (Handford et al. 1991). Thus, when taken together, EGFL7 is a relatively small, modular assembled protein composed of multiple domains that undergoes several post-translation modifications and is ultimately secreted from expressing cells.

1.2 EGFL7 function in health and disease

Under physiological conditions, EGFL7 plays important roles in embryonic development, vascular homeostasis, and tissue maintenance. Regarding expression, EGFL7 is predominantly associated with the proliferating endothelium, with high levels observed during embryogenesis and organogenesis. For example, EGFL7 is shown to be involved in vascular tubulogenesis in zebrafish embryos, with knockdown disrupting this process (De Mazière et al. 2008; Parker et al. 2004). In addition, knockdown of EGFL7 in mice disrupts feto-placental vascularization (Lacko et al. 2017). In adult tissues, EGFL7 expression is significantly lower around non-proliferating vessels compared to the developing embryo (Campagnolo et al. 2005; Parker et al. 2004), with increased levels of EGFL7 observed in the proliferating, migrating, or remodeling endothelium in adults, for example, in tissues following vascular injury or during pregnancy (Campagnolo et al. 2005). EGFL7 is shown to influence smooth muscle cell migration for both blood and lymph vessels (Mahamud et al. 2019; Soncin et al. 2003), while loss of EGFL7 in human microvessels suppresses VEGF-A-induced angiogenic sprouting. This leads to an overproduction of endothelial filopodia, reduces collagen IV deposition, and impairs the barrier function of endothelial cells (Usuba et al. 2019). At a molecular level, EGFL7-mediated angiogenesis involves the Notch signaling pathways as a non-canonical ligand (Bicker and Schmidt 2010; Nichol and Stuhlmann 2012), as well as interactions with cell surface integrins, in particular α_vβ₃ and α₅β₁ (Chim et al. 2015; Dudvarski Stankovic et al. 2018; Nikolić et al. 2013).

Beyond vascular development and homeostasis, EGFL7 is involved in the regulation of the central nervous system, increasing in cells of the maturing brain and remaining at high levels into old age (Bicker and Schmidt 2010). Moreover, the source of central nervous system EGFL7 were shown to be endothelial cells and neurons. Data suggest that this protein is an endogenous inhibitor of Notch signaling in the adult brain, e.g., regulating neural stem cell proliferation and differentiation. Recent work has demonstrated that the expression of EGFL7 in the subventricular zone functions as an inhibitor of Notch signaling, which decreases the self-renewal capacity and proliferation of neural stem cells, shifting differentiation patterns towards an excess of oligodendrocytes and neurons (Schmidt et al. 2009). Additional work in the nervous system of mice has revealed that loss of EGFL7 induces an accumulation of activated neural stem cells in the subventricular zone, causes quiescence of activated neural stem cells, and promotes neuronal progeny towards differentiation via Dll4-induced Notch signaling at the blood vessel-stem cell interface (Bicker et al. 2017). This alteration in neural stem cells activity ultimately leads to a decrease in inhibitory neurons in the olfactory bulb which causes a profound disruption of neuronal network synchronicity, causing severe physiological deficits in olfaction. Interestingly, Barth and colleagues recently reported that EGFL7 is secreted by neural stem cells, neural precursor cells, and mature neurons in the other significant neurogenic niche of the adult brain, the subgranular zone of the hippocampus (Barth et al. 2023). They demonstrate that loss of EGFL7 correlates with an increased expression of VEGF-D and upregulation of neurogenesis. Behaviorally, mice with EGFL7 ablated in the hippocampus were shown to demonstrate increased spatial learning and memory. Taken together, growing evidence thus indicates that EGFL7 is an important regulator of neurogenesis and nervous system health.

Along with developmental and physiological functions, EGFL7 is also implicated in various human diseases. One of the most well studied disease contexts involving EGFL7 is human cancer. Tumor growth and metastasis often rely on the formation of new blood vessels to provide nutrients and oxygen to the growing tumor mass. Previous work has demonstrated that EGFL7 is involved in this process, facilitating tumor metastasis, proliferation, and angiogenesis (de Oliveira et al. 2023). Specifically, a marked increase in the expression of EGFL7 in the tumor microenvironment has been reported for multiple tumor types, including laryngeal (Wang et al. 2015) ovarian (Oh et al. 2014), breast (Philippin-Lauridant et al. 2013), kidney (Zhai et al. 2019), brain (Huang et al. 2010; Wang et al. 2017), colorectal (Juan et al. 2021), and gastric (Deng et al. 2016) cancers. Moreover, several investigations indicate that this increase in EGFL7 expression may provide information relevant to cancer prognosis (Hansen et al. 2017) and treatment responsiveness (Hansen et al. 2013, 2014). Mechanistically, increased EGFL7 expression is associated with the ERFR-PI3K-Akt (Huang et al. 2014; Juan et al. 2021; Luo et al. 2014) and NOTCH (Bill et al. 2020; Tang et al. 2019; J. Wang et al. 2017; Y. Wang et al. 2022) pathways, facilitating tumor angiogenesis and cancer cell metastasis (Xu et al. 2014). Previous work also indicates that regulation of adhesion molecule expression on the endothelial cell surface via EGFL7 is driven by nuclear factor kappa B (NF-κB) signaling, with EGFL7 blocking the proteasomal degradation of the NF-κB-inhibitor IκBα, subsequently reducing adhesion molecule expression (Pinte et al. 2016). In gastric cancer cells, EGFL7 has further been linked to downregulation of E-cadherin, triggering epithelial-to-mesenchymal-transition and consequentially facilitating metastasis (Luo et al. 2014). In glioma, EGFL7 was shown to be secreted by the tumor blood vessels (but not the tumor cells), which increases endothelial integrin α₅β₁ surface expression and promoting fibronectin-stimulated angiogenic sprouting (Dudvarski Stankovic et al. 2018).

Growing evidence thus indicates that EGFL7 is a promising target for further studies regarding cancer prognosis and treatment. Indeed, in metastatic colorectal cancer, EGFL7 was shown to be a biomarker in patients receiving combination of bevacizumab and chemotherapy, with a significant relationship between response rates and EGFL7 along with lower EGFL7 expression at the invasive front in responding patients (Hansen et al. 2014). In addition, promising evidence suggests that anti-EGFL7 antibody therapy in combination with anti-VEGF and chemotherapy could enhance stress-induced endothelial cell death, limiting progression and prolonging cancer patient survival (Johnson et al. 2013). We note however that phase II clinical trials utilizing parsatuzumab (an anti-EGFL7-antibody) together with bevacizumab (an anti-VEGF antibody) and chemotherapy in both small cell lung cancer (García-Carbonero et al. 2017) and colorectal cancer (von Pawel et al. 2018) were to date unable to show significant benefits compared to standard care. Importantly however, successful targeting of EGFL7 in cancer may depend on the tumor type. For example, treatment may be more beneficial for patients in other cancer types such as malignant brain tumors given the strong induction of angiogenesis in brain tumors or their growth in an immune-privileged area of the brain (Dudvarski Stankovic et al. 2018; Huang et al. 2010). Thus, when taken together, EGFL7 is a molecule demonstrated to be upregulated in a wide range of cancers, implicated in tumor angiogenesis, and thus representing an important emerging biomarker and target in cancer therapy. Ongoing investigations will benefit from elucidating specific mechanisms across different tumor types in order to improve the efficacy of anti-EGFL7 cancer therapy.

Beyond this role cancer, EGFL7 is also implicated in various other pathologies including stroke (Nikolić et al. 2013), preeclampsia (Lacko et al. 2014) and inflammatory disease (Pinte et al. 2016). In addition, recent work has demonstrated that EGFL7 may be an interesting target for multiple sclerosis (Larochelle et al. 2018; Uphaus et al. 2018). Specifically, Larochelle and colleagues demonstrated that the expression of EGFL7 is increased in the CNS vasculature of patients with multiple sclerosis, as well as in mice with experimental autoimmune encephalomyelitis (Larochelle et al. 2018). Furthermore, perivascular CD4 T lymphocytes were found to be in close proximity to ECM-bound EGFL7 within multiple sclerosis lesions, while activated T lymphocytes in both humans and mice demonstrated an increase in the EGFL7 ligand α_vβ₃ integrin expression. EGFL7-knockout mice exhibited an earlier onset of experimental autoimmune encephalomyelitis and heightened infiltration of T lymphocytes into the brain and spinal cord parenchyma, with the administration of recombinant EGFL7 ameliorating the experimental autoimmune encephalomyelitis, reducing expression of MCAM, and enhancing blood-brain barrier integrity in the EGFL7-knockout mice. These findings thus suggest a role for EGFL7 in limiting central nervous system immune infiltration, which could potentially prove a promising target for multiple sclerosis therapy.

Therefore, when taken together, the current body of evidence indicates that EGFL7 is a secreted protein, which facilitates a permissive microenvironment for blood vessel formation and growth, as well as regulating cell behavior in the adult central nervous system (see Figure 1). Importantly, based on involvement in tumor metastasis, proliferation, and angiogenesis, as well as various other pathologies, EGFL7 represents an important potential target for clinical therapeutics.

Figure 1:

EGFL7: gene and protein structure, as well as function. Egfl7 consists of 11 introns and exons, with two alternative exons, 1a and 1b. The open reading frame begins at exon 3, continuing along the rest of the gene. Further, egfl7 encodes for microRNA-126, which is embedded within intron 7. Binding sites for the transcription factors ETS-related gene (Erg) and GATA2 are present at the 5′ end. The domain structure at the N-terminus of the EGFL7 protein includes a secretion signal peptide (SSP), facilitating cell secretion and an Emilin-like domain (EMI). Located centrally, two epithelial growth factor (EGF)-like domains are present with the more C terminal one relying on Ca²⁺ binding for proper folding. Among others, EGFL7 interacts with NOTCH receptors, integrins (e.g., α_vβ₃ and α₅β₁ motif) and extracellular matrix proteins. Egfl7 is functionally involved in vasculogenesis, angiogenesis and neurogenesis under physiological conditions, while also being implicated in tumor angiogenesis and neuroinflammation in pathological conditions. Created with BioRender.com.

2.1 E. coli expression system

The use of E. coli bacteria for expressing recombinant human and mouse proteins is a popular choice due to its ease of use, rapid growth, and cost-effectiveness. Indeed, the earliest protocol by Caetano and colleagues describing the expression of recombinant EGFL7 utilized E. coli (Caetano et al. 2006) and a handful of studies have applied this expression system to investigate EGFL7’s function (Delfortrie et al. 2011; Lelièvre et al. 2008; Philippin-Lauridant et al. 2013). However, one drawback of isolating secreted mammalian proteins from bacteria is the difficulty in establishing proper secondary modifications. Another serious issue with using E. coli as the recombinant EGFL7 expression system is the necessity to recover the purified protein from inclusion bodies and the subsequent need to apply buffers allowing for correct post-translational folding (Caetano et al. 2006; Picuric et al. 2009). Specifically, Caetano and colleagues expressed and purified mouse EGFL7 from E. coli in the presence of urea to facilitate protein purification and extraction. However, due to the denaturing effects of urea, the EGFL7 protein required additional refolding. While Caetano and colleagues were able to successfully refold the protein and validate biological activity via the repression of human aortic smooth muscle cell migration induced by platelet-derived growth factor-BB (PDGF-B) (Soncin et al. 2003), they had difficulties establishing a buffer solution which would allow the protein to remain soluble at relatively high concentrations. After attempts with different pH, salts, organic solvents, and detergents, the only successful means of stabilizing folded, active EGFL7 in solution was in the presence of a high concentration D-arginine, in the absence of urea. Unfortunately, this high arginine concentration in the buffer solution creates a high osmotic pressure and has the potential to interfere with nitric oxide formation and signaling, thus making the purified protein most likely unusable for both in vitro and in vivo experimentation (Oess et al. 2006; Picuric et al. 2009). Our experience with recombinant EGFL7 purified from E. coli is that the protein can be applied to coat cell culture dishes and was successfully used in adhesion assays. However, to date, our attempts to apply recombinant EGFL7 from E. coli for the inhibition of PDGF-B-induced cell migration described above have failed, limiting its experimental use. Additionally, it is important to note that post-translational modifications (such as phosphorylation and glycosylation) are absent in this expression system, further limiting application. Thus, while the E. coli expression system may be useful in certain circumstances (e.g., testing cell adhesion or antibody affinity), these limitations ultimately makes this system most likely ill-suited for laboratory studies investigating the broader physiological function of EGFL7, in particular in vivo.

2.2 HEK293 and other mammalian cell expression systems

At face value, the epithelial cell line HEK293 or other mammalian cells would be ideal expression systems for active mouse or human EGFL7 production as they offer correct post-translational secondary modifications. At the least, mammalian cells expression systems come closer to the original protein than the bacterial recombinant protein, particularly due to post-translational modifications more akin to those of the protein’s native state. However, both we and others have experienced practical limitations with mammalian cells, reducing their usefulness at producing sufficient, usable yields of active recombinant EGFL7 for experimental purposes. Specifically, Schmidt and colleagues reported that they were unable to detect the EGFL7 protein in the condition medium of HEK293 and Chinese Hamster Ovary (CHO) cells, despite the cells overexpressed the full-length protein (Schmidt et al. 2007a). However, they were able to detect low levels of EGFL7 when overexpressed in fibroblast cell lines, such as chicken embryonic fibroblasts (CEF) and NIH3T3 cells. Moreover, staining for fibroblast-secreted EGFL7 revealed that the protein is located outside the cell and becomes more and more fibrous over time. They speculate that this is observed in fibroblasts compared to HEK293 or Chinese Hamster Ovary cells because of fibroblast’s role in ECM deposition, with EGFL7 being secreted from these cells and integrated into the ECM (Schmidt et al. 2007b). In other words, this evidence suggests that EGFL7 is not a classic diffusible factor but is rather an ECM component. This is further corroborated by the fact that extracellular EGFL7 significantly overlaps fibronectin when endogenously expressed by human umbilical vein endothelial cells (Schmidt et al. 2007a).

Our experience in using HEK293 cells to produce recombinant EGFL7 reflects these reported issues. Specifically, we were successfully able to detect high levels of recombinant EGFL7 utilizing HEK293 cells. However, the protein spontaneously and instantly precipitated upon secretion, profoundly limiting our ability to obtain a usable yield. Put another way, our experience with mammalian cells suggests that the higher the concertation of the purified protein, the greater the probability of precipitation. Ultimately, consistent issues with protein stability and yield may potentially limit the use of mammalian cells as viable expression vectors for recombinant EGFL7 production. Thus, while mammalian and indeed human cells would be the preferred system for expressing recombinant EGFL7, currently no successful protocol using such cells is described in the literature.

2.3 Baculovirus-based Sf9 insect cell expression

The baculovirus system is widely used for expressing recombinant human proteins (Jarvis 2009). Utilizing this expression system, Picuric and colleagues describe a protocol involving the expression and purification of active recombinant EGFL7 (Picuric et al. 2009). Specifically, Picuric and colleagues expressed human EGFL7 in the Sf9 insect cell line in an insect condition medium, purifying the protein from the cell lysate by immobilized metal affinity chromatography and gel filtration. Of note, EGFL7 was only detectable in the total cell lysate. Urea was not required throughout this process. The conformation of EGFL7 was established by circular dichroism spectrometry, which indicated that the folding of the protein correlated with theoretically calculated values of the native EGFL7 protein. Lastly, they demonstrated the biological activity of the protein via the repression of PDGF-B-induced migration of 10T1/2 myofibroblast cells, in analogy to the assay previously used as biological activity control and applied to smooth muscle cells (Soncin et al. 2003). Importantly, the buffer solution required to maintain this recombinant EGFL7 in solution did not require arginine in this protocol. Thus, Picuric and colleagues described a procedure that overcomes the issues with the arginine-containing buffer solution present with the E. coli expression system. Moreover, they were able to isolate sufficient quantities of recombinant EGFL7 to make this method viable for physiological investigations. Indeed, we and others have successfully used this expression system to investigate the function of EGFL7, both in vitro and in vivo (Larochelle et al. 2018; Salama et al. 2017; Schmidt et al. 2009). For example, we have utilized insect cell-derived recombinant EGFL7 to demonstrate that EGFL7 modulates Notch signaling and affects neural stem cell renewal (Schmidt et al. 2009).

Also of note, the use of the baculovirus-based Sf9 insect cell expression system has also been reported for a related protein within the EGFL family, EGFL6. Specifically, Oberauer and colleagues describe the production of recombinant EGFL6 via the baculovirus-based Sf9 insect cell expression system in a manner similar to recombinant EGFL7, including testing activity via a cell adhesion assay (in this case using human adipocytes). The main findings of this study demonstrates that the EGF-repeats of EGFL6 mediate binding to the cellular surface, as well as enhancing the proliferation, of human adipocytes. The authors also indicate that the EGFL6 detected in the conditioned media of human adipocytes was larger (∼65 kDa) than the recombinant protein and they suggest this may be due to differences in post-translational modifications occurring during recombinant protein production. We note that such differences in post-translational modifications may also be the case for recombinant EGFL7 and this should be taken into account when deciding on a preferred expression system.

Based on this, the baculovirus-based Sf9 insect cell expression system offers several advantages over and above the E. coli and mammalian cell expression systems. First, as described in the published protocol (Picuric et al. 2009), this expression system does not require the buffer solution to contain arginine, reducing concerns regarding disruptive, potentially biologically relevant cross-reactivity. Second, when compared to the E. coli expression system for EGFL7, baculovirus-based Sf9 insect cell expression of EGFL7 has the advantage that post-translational modifications are present, although importantly these are most probably simpler than in mammalian cells. Third, this system allows for sufficient yield and stability for experimental purposes. Disadvantages of the baculovirus-based Sf9 insect cell expression system are first and foremost the simpler post-translational modifications that are most likely present, as these may influence the recombinant proteins function compared to the native form. Second, from a practical perspective the baculovirus-based Sf9 insect cell expression system is more expensive and time consuming than the E. coli expression system. Taken together, depending on experimental goals, the baculovirus-based Sf9 insect cell expression system enables the production of active recombinant EGFL7 in its native conformation, including simple post-translational modifications, along with ensuring adequate yield and stability for experimental applications.