GV21 ultrasound imaging tracks VEGFR2 for early tumor therapy assessment

Ultrasound imaging has long been prized for its safety, accessibility, and real-time capability, but a new generation of molecular contrast agents is pushing the technology far beyond simple anatomy. Researchers have now developed an engineered nanoscale ultrasound contrast agent, dubbed GV21, that can visualize vascular endothelial growth factor receptor 2 (VEGFR2) not only on tumor blood vessels but also on the tumor cells themselves, a feat that conventional microbubble agents cannot achieve. By combining this nanoscale probe with a clever signal-subtraction strategy, the team demonstrated a way to noninvasively measure how much of a tumor’s VEGFR2 payload lies beyond the vasculature, and to detect the earliest molecular signs of response to targeted cancer therapy days before tumor shrinkage becomes measurable.

The work, led by Xiaoxin Liang, Shilin Lu, Jianhua Zhou, and Fei Yan of the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, addresses a fundamental physical limitation of ultrasound molecular imaging (USMI). The technique relies on targeted microbubbles—gas-filled, lipid-shelled particles typically one to ten micrometers in diameter—that resonate under ultrasound pulses and bind to specific molecular markers on blood vessel walls. The most advanced clinically translatable example is BR55, a VEGFR2-targeted microbubble that has performed well in preclinical models of pancreatic, breast, prostate, and colon cancer and has progressed into Phase 2 clinical trials, with first-in-human studies demonstrating safety and feasibility in prostate, breast, and ovarian lesions. But the very size that gives microbubbles their strong acoustic signal also confines them strictly within the bloodstream. Because they cannot squeeze through the leaky but still restrictive tumor vasculature, microbubble-based imaging essentially reports only VEGFR2 expressed on vascular endothelium, potentially missing a large and clinically important portion of the tumor’s molecular profile.

VEGFR2 is best known as a driver of angiogenesis, the process by which tumors recruit new blood vessels, but it is increasingly recognized that many tumor cell types themselves overexpress the receptor. Studies across lung, glioma, prostate, and squamous cell carcinoma models have linked tumor cell VEGFR2 expression to aggressive behavior, autocrine survival signaling, and poor prognosis. A complete picture of VEGFR2 biology within the tumor microenvironment therefore requires interrogating both the vascular and the tumor cell compartments simultaneously—something microbubbles, trapped inside blood vessels, cannot do alone.

The solution came from an unlikely biological source: the halophilic archaeon Halobacterium sp. NRC-1. This microorganism produces biosynthetic gas vesicles—hollow, protein-shelled nanostructures approximately 200 nanometers in diameter—that scatter ultrasound effectively enough to be visible on standard clinical scanners. Because of their nanoscale dimensions, these gas vesicles can exit the tumor vasculature through endothelial gaps and penetrate the interstitial tumor tissue, placing them in direct contact with tumor cells. To exploit this property, the research team engineered V21, a camelid-derived single-domain antibody (nanobody) that binds VEGFR2, and covalently attached it to the protein shells of purified gas vesicles using a maleimide-thiol conjugation chemistry, with NHS-PEG2000-maleimide serving as the linker. The resulting probe, GV21, carries roughly 800 V21 molecules per particle. A parallel preparation of conventional lipid microbubbles decorated with the same nanobody—designed to mimic BR55—served as the intravascular comparator, with an estimated 38,000 nanobody molecules per microbubble.

Characterization confirmed that the two agents occupied sharply different physical regimes. Dynamic light scattering measured the gas vesicles at roughly 210 nanometers in diameter, while the microbubbles measured approximately two micrometers. To produce comparable ultrasound contrast on a clinical Resona 9 scanner operating in nonlinear contrast mode, gas vesicles had to be administered at concentrations roughly four orders of magnitude higher than microbubbles—around 10¹² particles per milliliter versus 10⁸—because each individual nanobubble generates a much weaker acoustic signal than each microbubble. In vitro binding experiments showed that GV21 attached strongly to VEGFR2-positive cells, with fluorescence signal ratios roughly 30- to 40-fold higher than control gas vesicles on VEGFR2-positive tumor and endothelial cells, while showing essentially no binding to VEGFR2-negative cells. In flow chamber assays simulating vascular shear stress, GV21 outperformed BR55-mimic microbubbles at every shear rate tested, adhering at roughly twice the rate even at physiologically relevant wall shear rates of 150 s⁻¹, likely because the smaller, denser nanoparticles are less prone to detachment under flow.

The decisive experiments took place in mice bearing human colon cancer xenografts. The team selected SW620 cells as a VEGFR2-positive model, with more than 99 percent of cells expressing the receptor, and HT29 cells as a VEGFR2-negative comparator with only about 3 percent positivity. After intravenous injection, GV21 produced significantly stronger and more sustained tumor signals than the BR55-mimic in SW620 tumors—74.27 versus 41.53 arbitrary units at 10 minutes post-injection—consistent with its ability to bind both vascular and extravascular VEGFR2. In HT29 tumors, where VEGFR2 is confined to the vasculature, the two probes generated statistically indistinguishable signals. Fluorescence microscopy of excised tumors confirmed the mechanism: BR55-mimic particles remained confined to the CD31-positive vascular endothelium, while GV21 particles appeared both along vessels and deep within extravascular tumor tissue.

The subtraction strategy then delivered its key result. By numerically subtracting the BR55-mimic signal from the GV21 signal within the same tumor region of interest, the researchers isolated the extravascular, tumor cell-associated VEGFR2 contribution. In SW620 tumors this subtracted signal averaged 37.65 arbitrary units, compared with just 9.36 in HT29 tumors—a highly significant difference that matched histological measurements of VEGFR2 in CD31-negative tumor regions. Correlation analysis was striking: the subtracted GV21-minus-BR55 signal correlated strongly with histologically quantified tumor cell VEGFR2 expression (R² = 0.86), while the GV21 signal alone tracked total tumor VEGFR2 (R² = 0.77), and the BR55-mimic signal tracked vascular VEGFR2 within CD31-positive regions (R² = 0.61). In effect, the two probes together provide a noninvasive compartmental analysis of receptor expression that previously required tissue biopsy.

The clinical payoff emerged in the therapy monitoring experiments. The researchers treated SW620 tumor-bearing mice with apatinib, an orally available selective inhibitor of the VEGFR2 tyrosine kinase approved in China for gastric cancer, at 50 milligrams per kilogram daily for 14 days, and imaged at baseline and on days 1, 3, and 7. Remarkably, within just 24 hours—before any measurable change in tumor volume—the GV21 signal had plummeted from 80.63 to 39.60 arbitrary units, while the BR55-mimic signal declined only modestly and did not reach statistical significance. The contrast makes biological sense: apatinib rapidly suppresses VEGFR2 phosphorylation and downstream AKT signaling in tumor cells, a molecular change that Western blotting of tumor tissue confirmed within 24 hours, whereas the vascular endothelial receptor pool is comparatively stable in the earliest phase of anti-angiogenic therapy, and structural vascular regression takes days to become visible. Immunohistochemistry corroborated the imaging: tumor cell VEGFR2 fluorescence dropped by roughly 80 percent by day 1, while CD31-positive vascular area only began to shrink significantly by day 3 and fell further by day 7.

To track the tumor cell compartment specifically over time, the team computed the change in each probe’s signal relative to baseline and then subtracted the microbubble-derived change from the gas vesicle-derived change. This ΔGVs–ΔMBs metric remained significantly elevated throughout days 1, 3, and 7 of treatment, indicating a persistent suppression of tumor cell-associated VEGFR2 lasting at least a week, while control mice receiving phosphate-buffered saline showed no signal changes at any time point. An alternative calculation sequence—first subtracting probe signals at each time point, then comparing to baseline—produced equivalent results, demonstrating that the analytic framework is robust to methodological choice.

Safety testing supported the translational promise of the approach. Neither GV21 nor BR55-mimic affected cell viability in vitro at the concentrations tested, neither induced hemolysis of red blood cells, and intravenous injection in mice produced no abnormalities in liver and kidney serum biomarkers at one and seven days, with heart, lung, liver, kidney, and spleen histology remaining normal on hematoxylin and eosin staining.

The implications extend beyond colon cancer. Because VEGFR2 is implicated in tumor aggressiveness across multiple cancer types, and because anti-angiogenic therapies are frequently limited by the absence of early, noninvasive efficacy readouts, a dual-probe USMI framework that resolves vascular versus tumor cell VEGFR2 could accelerate dose selection and patient stratification. The authors acknowledge remaining hurdles: the subtraction approach is an indirect approximation, the study covered a limited set of tumor models, and clinical translation of biosynthetic gas vesicles will require scalable manufacturing, batch consistency, and long-term pharmacokinetic and immunogenicity evaluation. Still, the demonstration that nanoscale acoustic biosensors can peer beyond the blood vessel wall—and that pairing them with conventional microbubbles can mathematically separate two molecular compartments invisible to each probe alone—marks a significant conceptual advance for molecular ultrasound imaging, and offers an early-warning system for targeted cancer therapy that could one day spare patients weeks of ineffective treatment.