High-quantity capture of heterogeneous circulating tumor cells in vivo based on a multinanobody-functionalized carboxymethyl cellulose cryogel

Abstract

Circulating tumor cells (CTCs) carry multidimensional information from primary tumors, yet current separation strategies lead to insufficient quantity of cells available for robust downstream analyses. To address this issue, we propose an in-vivo capture strategy for high-quantity CTC enrichment based on a multinanobody-functionalized carboxymethyl cellulose cryogel (MNB-PAAC_cryo). By addressing the trade-off between pore size and mechanical property, the MNB-PAAC_cryo sets up a sieving barrier in the bloodstream. It allows free passage of blood cells through its interconnected macropores, while enhancing CTC-nanobody interactions, contributing to a high CTC capture efficiency of 72.3 ± 4.5% in the bloodstream. Consequently, up to 2.7 × 10³ CTCs could be harvested from 4000 mL blood at 1 CTC/mL using 0.1 cm³ of the MNB-PAAC_cryo. In vivo experiments on cancer rat models demonstrated that the MNB-PAAC_cryo could capture diverse CTC subtypes across multiple cancers, and the captured CTCs exhibited consistent information with the primary tumor tissues. With its robust capability for high-quantity CTC enrichment, the proposed strategy will not only improve the reliability of CTC-based cancer analysis, but may also facilitate the establishment of patient-derived tumor organoid, showing great potential to accelerate the development of personalized cancer medicine.

Graphical abstract

1. Introduction

Circulating tumor cells (CTCs) are cancer cells shed from primary tumors into the bloodstream that carry intact information of the solid tumor [[1], [2], [3]]. They enable the cytological and molecular diagnosis to assess tumor burden, prognosis and treatment response [4,5], showing great potential in cancer detection, diagnosis and treatment, especially in early cancer detection, difficult-to-biopsy tumors, and real-time monitoring of tumor dynamics [6,7]. CTCs have been well recognized as a promising cancer biomarker since its first identification in 1869 [8], and a variety of technologies have now emerged for CTC-based cancer diagnosis, including CTCs enumeration-based cancer detection [3], immunohistochemistry-based phenotypic analysis [[9], [10], [11]], and single-cell multi-omics [12,13] (see Scheme 1).

Scheme 1.

Nanobody-functionalized carboxymethyl cellulose cryogel toward high-quantity capture of circulating tumor cells across multiple cancer types from the bloodstream for cancer diagnosis.

However, CTC-based liquid biopsies have yet to be widely accepted as a routine diagnostic protocol by the medical community, owing to its controversial performance in clinical cancer diagnose. The most critical issue lies in the statistical unreliability caused by the insufficient CTC number for analysis [14,15], mainly due to the rarity of CTCs in circulating blood (typically 1-10 CTCs per mL [16,17]), the limited volume of clinical blood sampling (<10 mL for a single blood draw), as well as the capture efficiency issue caused by the interference from hematologic cells [18]. The extremely small sample size fundamentally undermines the statistical confidence of any subsequent analysis. Significant analytical errors frequently occurred, resulting in unacceptably high false-negative rate in cancer detection. Moreover, given the inherent heterogeneity of CTCs [[19], [20], [21], [22], [23]], the limited number of CTCs fails to capture the overall tumor landscape, resulting in unrepresentative and potentially misleading findings. A high CTC quantity is essential not only to ensure analytical reliability but, more importantly [24,25], to enable downstream analysis such as cytologic, phenotypic, and molecular diagnosis [26], which is much more crucial for advancing CTCs from dispensable prognostic biomarkers towards reliable diagnostic tools that can provide comprehensive cancer information in clinical oncology.

Theoretically, processing large blood volumes (hundreds to thousands of millilitres) could substantially increase the CTC yield. However, the volume of clinical blood collection is restricted by safety considerations, typically to less than 50 mL [27,28]. To circumvent this limitation, in vivo CTC capture strategies have been proposed, whereby functionalized capture materials are implanted into the circulatory system. Such strategies enable the entire circulating blood volume (more than 4000 mL for an adult) to interact with the capture surface, thereby greatly enhancing CTC yield. Besides, normal blood components such as blood cells and plasma proteins can be returned to the patients, avoiding the blood loss associated with large-volume sampling. However, currently, such strategy turns out to be a Pyrrhic victory. Despite its conceptual appeal, the actual number of captured CTCs remains very low as capture efficiency issue occurs when it comes to in vivo CTC capture. While significant progress has been made in in vitro CTC capture using biological or physical isolation technologies [[29], [30], [31], [32], [33], [34], [35], [36]], with reported efficiencies reaching 90% or higher [[37], [38], [39], [40]], the situation is markedly different in vivo. When these strategies are applied in the bloodstream, capture efficiencies drop dramatically to below 10% [41]. The representative in vivo CTC capture tool, CellCollector, exhibits a capture efficiency of merely 0.0016 ± 0.0003% [42]. MagWIRE, which employs the immunomagnetic bead approach for in vivo applications, attains a capture efficiency of only 1-8% even at 100 CTC/mL [43]. In the hemodynamic environment of the bloodstream, the high flow velocity and shear forces greatly reduce the collision frequency and contact time between CTCs and the capture surfaces [44,45]. Besides, unlike in vitro approaches that can employ pretreatments such as erythrocyte lysis and leukocyte depletion on demand to enhance capture efficiency and specificity [46], in vivo strategies should capture CTCs in situ directly from the bloodstream without any pretreatments. The strong interference from the complex blood components with high abundance poses additional challenges to achieving efficient and specific CTCs capture [47,48]. Efforts should be made to improve in vivo capture efficiency and harvest far more CTCs to address the statistical reliability issue of CTC-based liquid biopsy.

To address the aforementioned in vivo capture bottlenecks dominated by severe hemodynamic shear force and insufficient CTCs contact under blood flow, we developed a robust strategy toward high-quantity enrichment of CTCs from the bloodstream based on a well-designed cryogel, which was expected to possess interconnected macropores allowing free passage of blood cells, while enhancing CTCs interactions via structure-induced cell deceleration to improve the CTCs capture efficiency. To fulfill these design expectations with adequate structural integrity, carboxymethyl cellulose was selected as the skeleton of the cryogel to remarkably improve its mechanical robustness. It is based on the hypothesis that the dense intermolecular and intramolecular hydrogen bonding within carboxymethyl cellulose, as well as the chain rigidity resulting from the high rotational energy barrier of its polysaccharide backbone, collectively could endow the cryogel with excellent mechanical properties. Benefiting from such excellent mechanical merits, the cryogel would exhibit sufficient mechanical strength to withstand harsh hemodynamic loading for stable long-term in vivo application. When functionalized with specific nanobodies targeting canonical cancer biomarkers such as EGFR, EpCAM and Vimentin, the cryogel is expected to set up a sieving barrier in the bloodstream, which could allow free passage of blood cells through its interconnected macropores, while enhancing CTC-nanobody interactions, so that a high-quantity capture of heterogeneous CTCs in vivo might be achieved.

2. Materials and methods

2.1. Materials

Sodium carboxymethyl cellulose (CMC, M_w = 778 kDa, Ð = 2.4, apparent values obtained from GPC calibrated with dextran standards; carboxymethyl group substitution degree: 0.9) was purchased from Sigma-Aldrich (Shanghai) Trading Co., Ltd. (Product No. 419303), N,N′-Methylenebisacrylamide (MBA), poly(ethylene glycol) diacrylate (Mn = 6000, PEGDA6000), trimethylolpropane triacrylate (TMPTA), polyacrylic acid (PAA, Mn = 100,000), 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), allylamine (AA), allyl glycidyl ether (AGE), N,N,N′,N′-tetramethylethylenediamine (TEMED), ammonium persulfate (APS), and acrylamide (AAM) were obtained from Shanghai Aladdin Biochemical Technology Co., Ltd. 2-(N-Morpholino) ethanesulfonic acid (MES) was sourced from Beijing Solarbio Science & Technology Co., Ltd. Anti-EGFR nanobody (NB_EGFR), anti-EpCAM nanobody (NB_EpCAM), and anti-Vimentin nanobody (NB_VIM) were gifts from Kangyuan Biomedical Tech. (Dalian) Co., Ltd, prepared by prokaryotic expression. DAPI dye and rhodamine-phalloidin were procured from Saiwen Innovation (Beijing) Biological Technology Co.,Ltd.

2.2. Synthesis of VCMC

Sodium CMC (1.0 g, containing 4.2 mmol of carboxyl groups) was dissolved in 100 mL of MES buffer (100 mM MES, 50 mM NaCl, pH 5.4). The carboxylic acid groups of CMC were then activated by adding EDC (0.81 g, 4.2 mmol) and NHS (0.48 g, 4.2 mmol) to the solution. After stirring for 5 min, allylamine (0.32 g, 5.6 mmol) was added, and the mixture was stirred at room temperature for 24 h (the process flow diagram is presented in Scheme S1a). Upon completion of the reaction, the product was precipitated by adding anhydrous ethanol (500 mL). The obtained vinylated carboxymethyl cellulose (VCMC) was purified by dialysis against deionized water (MWCO 3500 Da) for 3 days to remove unreacted allylamine and other small molecular impurities, followed by lyophilization to obtain a white fluffy solid.

2.3. Characterization of VCMC

For ¹H nuclear magnetic resonance (¹H NMR) spectroscopy, the VCMC (50 mg) and pristine CMC (50 mg) polymers were dried under vacuum and then dissolved in 450 μL of deuterium oxide (D₂O, with the residual H₂O signal at ∼4.7 ppm as the reference). The ¹H NMR spectra were recorded at 25 °C using a Bruker AVANCE III 500 spectrometer (Bruker, Switzerland). The obtained spectra were processed using standard baseline correction prior to analysis.

For Fourier-transform infrared (FTIR) spectroscopy, the dried samples were measured using an FTIR spectrometer (iN10, Thermo Fisher Scientific, USA) in the attenuated total reflectance (ATR) mode. The spectra were collected over a wavenumber range of 4000-500 cm⁻¹ with a resolution of 4 cm⁻¹ and 32 scans per sample at 25 °C. All spectra were baseline-corrected before analysis.

The weight-average molecular weight (M_w) of VCMC was measured by gel-permeation chromatography (GPC, 1260 Infinity Ⅱ, Agilent, USA). The sample was dissolved in 0.1 M NaNO₃ at 2.0 mg/mL, filtered through a 0.45 μm membrane, and 20 μL was injected. The mobile phase consisted of 0.1 M NaNO₃ at a flow rate of 1 mL/min; the column oven was maintained at 30 °C. Dextran standards (0.6, 1.5, 4, 15, 30, 43, 74, 129, 427.5, 1044, 1511 kDa) were used to construct the calibration curve by plotting log (M_w) versus retention time (R² > 0.99). Weight-average molecular weight, along with the dispersity (Ð = M_w/M_n), were calculated using the instrument software calibration.

2.4. Preparation and characterization of PAAC_cryo

Allyl glycidyl ether (AGE, 1 wt%), acrylamide (AAM, 3 wt%), and vinyl carboxymethyl cellulose (VCMC, 1 wt%, a degree of vinyl substitution of 0.6) were dissolved in deionized water (pre-boiled to remove dissolved oxygen). After cooling in an ice bath, N,N,N′,N′-tetramethylethylenediamine (TEMED, 0.08 wt%) and ammonium persulfate (APS, 0.08 wt%) were added. Following thorough stirring, the solution was promptly injected into a mold and cooled to −20 °C for 18 h, during which solvent crystallization generated ice templates, and the monomers and initiator became concentrated in the unfrozen microphase where polymerization proceeded. After subsequent thawing and washing thoroughly with deionized water, poly(AAM-co-AGE-co-VCMC) cryogel with interconnected macroporous structure was obtained (designated PAAC_cryo, the process flow diagram is presented in Fig. 1 and Scheme S1b). Scanning electron microscopy (SEM, Hitachi, ITACHI SU8010) was performed to characterize the morphology of the obtained cryogel: the samples were dehydrated with absolute ethanol, dried at 60 °C, sectioned, sputter-coated with gold, and then subjected to SEM observation.

Fig. 1.

Preparation and characterization of the PAAC_cryo cryogel. (a) Synthesis of the VCMC. (b)¹H NMR spectra of the VCMC. (c) Intramolecular and intermolecular hydrogen bonds in VCMC. (d) Diagram of the preparation of PAAC_cryo cryogel with interconnected macropores. (e) SEM images of the PAAC_cryo cryogel (prepared at −20 °C) and PAAC_hydro hydrogel (prepared at room temperature). (f) Fluid penetration performance of the PAAC_cryo and PAAC_hydro under gravity-driven and pressure-driven flow; the liquid was dyed in red. (g-h) Computational simulation of particle flow velocity (g) and particle flow trajectory (h) for PAAC_cryo and PAAC_hydro based on their reconstructed structural models. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

2.5. Preparation and characterization of NB-PAAC_cryo and MNB-PAAC_cryo

For the preparation of NB-PAAC_cryo, PAAC_cryo was immersed into NB_EGFR solution (1, 5, 10, 15 and 25 mg/mL, pH = 9.5) for 6 h to immobilize the nanobody onto the cryogel. Unconjugated nanobody was removed by washing thoroughly with phosphate buffered saline (PBS) buffer; and unreacted active sites were blocked by incubation in an ethanolamine solution overnight (4 wt% in PBS). After additional washing thoroughly with PBS, the NB_EGFR functionalized cryogel, NB-PAAC_cryo, was then obtained (the process flow diagram is presented in Scheme S1c). MNB-PAAC_cryo was prepared using a similar procedure in a mixed solution containing NB_EGFR (25 mg/mL), NB_EpCAM (25 mg/mL), and NB_VIM (25 mg/mL) nanobodies. The as-prepared cryogels were extensively washed by perfusion with a large volume of fresh PBS until no total organic carbon (TOC) was detectable in the washing solution. The successful synthesis of the final product was confirmed via scanning electron microscopy (SEM) and. Scanning electron microscopy (SEM) and X-ray diffraction (XRD, Shimadzu Corporation, XRD-6100) were performed for the characterization of the obtained cryogel.

2.6. CTCs capture of NB-PAAC_cryo under dynamic fluid condition

A closed-loop circulation circuit was established for the CTC capture assay under dynamic fluid condition. The system consisted of a peristaltic pump, an NB-PAAC_cryo cryogel column (0.1 cm³), and a cell suspension reservoir, connected using polyvinyl chloride tubing. For the capture experiments, a predetermined number of cells (A431, HeLa, HEK293, and HL60) were suspended in PBS. The cell suspension was circulated through the NB-PAAC_cryo cryogel column five times at a flow rate of 4 mL/min to perform the capture process. The capture efficiency (E_PBS) was determined by quantifying the change in cell concentration in the suspension before and after capture using fluorescence microscopy. Experimental details are provided in Supporting Information (Experimental Section S1.8.)

2.7. In vitro CTC capture of the NB-PAAC_cryo and MNB-PAAC_cryo in simulated bloodstream

The closed-loop circulation circuit used herein was identical to that described in Section 4.3, except that the CTCs were suspended in whole blood (from healthy adult donors) to simulate the bloodstream. For the CTC capture assay, A431 cells of different concentrations were suspended in 5 mL of human whole blood. The resulting CTC suspension was circulated through the NB-PAAC_cryo or MNB-PAAC_cryo column five times at a flow rate of 4 mL/min to conduct the capture process. Following capture, the NB-PAAC_cryo or MNB-PAAC_cryo cryogel was collected and processed into paraffin sections for staining. The number of CTCs captured by the cryogel was visualized and quantified according to the fluorescent microscopy images of the sections. Experimental details are provided in Supporting Information (Experimental Sections S1.9 and S1.10)

2.8. In vivo CTC capture from the bloodstream via extracorporeal circulation

The experiments were approved by the Animal Ethical Committee (No. DUTSBE260112-01) of Dalian University of Technology, and all the experiments were performed in GemPharmatech in accordance with the National Institutes of Health Guide for the care and use of laboratory animals (NIH Publications No. 8023, revised 1978).

Cancer models were established by subcutaneous inoculation of A431 (skin cancer), A549 (lung cancer), and HepG2 (liver cancer) cell lines into immunodeficient Sprague–Dawley (SD) rats. When the resulting tumors reached an approximate diameter of 25 mm, extracorporeal circulation was established for the in vivo CTC capture, shown as follows: The entire extracorporeal circulation circuit, consisting of a peristaltic pump and an MNB-PAAC_cryo cryogel column (0.1 cm³), was pretreated with heparin sodium solution (60 IU/mL) for 30 min (the heparin exposure would not compromise the CTC capture performance of the cryogel, Fig. S1). Indwelling needles were inserted into the cervical artery and vein for extracorporeal circulation device connection. Heparin was administered intravenously to the tumor-bearing rats for systemic anticoagulation at a dose of 240 IU/kg weight. After connecting the extracorporeal circuit to the cervical vessels, extracorporeal circulation was initiated for in vivo CTC capture at a blood flow rate of 2 mL/min for 2 h, with supplemental heparin sodium injections administered every 30 min (60 IU per injection). The tumor-bearing rats were anesthetized by isoflurane inhalation throughout the experiments. Following capture, the MNB-PAAC_cryo cryogel was collected and processed into paraffin sections for immunofluorescence staining using antibodies against a series of tumor biomarkers for microscopy. For quantification, each cryogel sample after the CTC capture experiments was sectioned into 100 consecutive slices. 10 slices were randomly selected for full-field cell counting to obtain the average count per slice, and the total captured cell number was then estimated based on the average count per slice and the total number of slices.

To assess the consistency of immunophenotypes between captured CTCs and primary tumour tissues, all imaging datasets were assigned random numeric identifiers by one researcher responsible for sample processing and immunofluorescence staining. Two independent analysts, who remained blinded to sample grouping and experimental metadata, performed CTC identification and immunophenotype evaluation relying exclusively on fluorescence signals. Personnel undertaking sample coding and staining were distinct from those performing image analysis to achieve double-blind assessment [49].

3. Results and discussion

3.1. Preparation of PAAC_cryo cryogel with interconnected macroporous structure and high mechanical strength

High flow velocity and shear forces in the bloodstream greatly reduce the collision frequency and contact time between CTCs and the capture surfaces, causing dramatic decrease in CTC capture efficiency. To address this issue, herein, we proposed a robust strategy toward high-quantity enrichment of CTCs from the bloodstream based on a well-designed cryogel. Cryogels are interconnected macroporous polymer networks formed at sub-zero temperatures [50,51], where ice crystals act as porogens that concentrate monomers in unfrozen regions for polymerization; subsequent melting yields a three-dimensional scaffold with interconnected macropores at microscale [52]. These interconnected macropores would ensure free passage of blood cells while increasing collision frequency and contact time with the material interface through structure-induced cell deceleration. When functionalized with specific ligand toward cancer biomarkers, such cryogels might enable highly efficient and selective capture of CTCs from the bloodstream.

However, the interconnected macroporous morphology is a double-edged sword, as it severely compromises the mechanical strength of cryogels [53]. Compressive deformation with pore collapse and blockage occurs under hemodynamic pressure, leading to the failure of their in vivo applications in CTC capture. To overcome the trade-off between interconnected macroporosity and mechanical strength, in this study, vinyl carboxymethyl cellulose (VCMC, Fig. 1a and b) was designed as a robust crosslinker to enhance the mechanical strength of cryogels. It is expected that the dense intermolecular and intramolecular hydrogen bonding within VCMC (Fig. 1c), coupled with inherent chain rigidity stemming from the high rotational energy barrier of its polysaccharide backbone [54], could contribute to sufficient mechanical strength of the cryogel to withstand harsh hemodynamic loading for stable in vivo applications.

Chemical characterizations demonstrated the successful synthesis of the VCMC. As shown in the ¹H NMR spectra of VCMC (Fig. 1b), the bands observed in the range of δ 5.0-6.0 ppm correspond to the characteristic resonance of vinyl protons within the allylamine segment in modified CMC, indicating the successful vinylation of the VCMC. The degree of vinylation substitution was estimated to be 0.6 (Fig. S2). Results of the FTIR spectroscopy (Fig. S3) further confirmed the vinylation. The retention of characteristic CMC absorption bands confirmed the integrity of the polysaccharide backbone, while the newly emerged peaks at ∼1640 cm⁻¹ (C=C stretching) and ∼1550 cm⁻¹ (amide N-H bending) demonstrated the formation of amide linkages between CMC and allylamine. The molecular weight of the VCMC was measured to be 835 kDa (Ð = 3.0) according to the GPC results (Fig. S4, apparent values calibrated with dextran standards).

As shown in Fig. 1d and S5-S9, among a series of crosslinkers with varying molecular chain length, number of crosslinking sites (N_cs), and backbone structure, the VCMC (M_w = 250K, N_cs = 577) crosslinked cryogel achieved the greatest enhancement in mechanical strength while maintaining their interconnected macroporous polymer networks. In this optimized cryogel, acrylamide (AAM, 3 wt%) was employed as the primary monomer to form the cryogel scaffold, allyl glycidyl ether (AGE, 1 wt%) provided reaction sites for immobilization of ligands targeting CTC biomarkers, while VCMC (1 wt%) enhanced the mechanical strength, obtaining the poly(AAM-co-AGE-co-VCMC) cryogel, abbreviated as PAAC_cryo (more information for the optimization could be found in the Supporting Information, Fig. S10–S12). The compressive modulus and compressive stress of the PAAC_cryo cryogel were 23.1 ± 3.2 kPa and 100 kPa (Fig. S13), respectively, and the fluid pressure remained below 10 kPa without any structural deformation at a linear velocity of 10 mm/s. The swelling ratio of MNB-PAACcryo was 0.3% after immersion in PBS solution for 24 h. Computational fluid dynamics (CFD) simulations revealed that the maximum lateral pressure exerted on the cryogel remained below 5 Pa throughout operation at a flow rate of 5 mL/min (Fig. S14).

SEM images showed that the PAAC_cryo possessed uniformly distributed, interconnected macropores ranging from 50 to 150 μm (Fig. 1e and S9). In contrast, no micron-scale pore structure was observed in its hydrogel form synthesized at room temperature (PAAC_hydro), indicating the ice-templating effect during cryogelation, where ice crystals act as porogens to generate the macroporous architecture of the PAAC_cryo. XRD analysis confirmed the amorphous nature of the cryogel, with lower crystallinity when compared with its hydrogel form (Fig. S15). Taking advantage of the interconnected macropores in PAAC_cryo, fluid transport through the cryogel was highly efficient. As shown in Fig. 1f, the PAAC_cryo cryogel allowed easy passage of liquid under both gravity and pressure without obvious deformation, whereas the PAAC_hydro hydrogel remained impermeable even under 1.5 MPa pressure, which instead caused its complete structural failure.

With the trade-off between pore size and mechanical strength of the PAAC_cryo cryogel being addressed, we investigated whether the interconnected macroporous structure could contribute to a structure-induced mass deceleration effect. As shown in Fig. 1g, CFD simulations demonstrated that the interconnected macroporous structure of the PAAC_cryo substantially decelerated CTC-mimicking particles compared to planar surfaces, with an approximately 460-fold increase in collision frequency, resulting in a much higher particle adsorption efficiency (Fig. 1h and S16). This phenomenon was further supported by real-time fluorescent particles tracing in the PAAC_cryo cryogel. As shown in Fig. S17, the fluorescent particles (diameter: 20 μm) migrated much more slowly through the PAAC_cryo cryogel (transit velocity: 0.01 mm/s) compared to the planar PAAC_hydro hydrogel (transit velocity: 3 mm/s). These results indicated that the interconnected macroporous structure of the PAAC_cryo cryogel could enhance the collision efficiency and contact duration with CTCs by structure-induced mass deceleration effect, which was anticipated to enhance the capture efficiency of CTCs in the bloodstream.

3.2. CTC capture of NB-PAAC_cryo in dynamic fluid condition

To capture the CTC, specific ligands targeting tumor biomarkers were immobilized onto the PAAC_cryo. A PAAC_cryo cryogel volume of 0.1 cm³ was used throughout this work, as it ensures sufficient CTC capture efficiency while minimizing material usage. Epidermal growth factor receptor (EGFR) [55], one of the most well-known biomarkers of tumor cells, was selected as a model herein, and its corresponding nanobody (NB_EGFR, K_D = 4.37 × 10⁻⁸ M) [56] was employed as the recognition unit to capture CTCs (Compared with conventional antibodies, nanobodies exhibit superior stability and tolerate subsequent chemical modification (Table S1, Figs. S18 and S19). The NB_EGFR was immobilized onto the PAAC_cryo through the ring open reaction between the epoxy groups on the cryogel and the amino groups on the nanobody, obtaining NB-PAAC_cryo(Fig. 2a). X-ray photoelectron spectroscopy (XPS) revealed a characteristic C-N⁺ peak at 400.7 eV in the N 1s spectrum after the nanobody immobilization (Fig. S20). This peak corresponds to the protonated amine species in the NB_EGFR (e.g., histidine, tryptophan, and asparagine), indicating the successful immobilization of NB_EGFR onto the cryogel. Fluorescence tracing further demonstrated the uniform spatial distribution of the NB_EGFR (FITC-labeled, green) throughout the cryogel. The maximum nanobody immobilization density for the PAAC_cryo was quantified to be 3.2 mg/cm³ (Fig. S21), which was adopted for all subsequent studies.

Fig. 2.

CTC capture of NB-PAAC_cryo in dynamic fluid condition. (a) Preparation of the NB-PAAC_cryo cryogel; and the fluorescence microscopy images of the NB-PAAC_cryo, NB_EGFR was FITC-labeled prior to immobilization in this assay for tracing purposes. (b) Schematic diagram of the circulatory system device for CTC capture. (c) Capture efficiencies of the NB-PAAC_cryo for A431, Hela, HEK293, and HL60 cells; the insert shows the EGFR expression levels of these cells. d) CTC capture efficiencies of the NB-PAAC_cryo and NB-SMs. (e) Fluorescence microscopy images of the CTCs captured by NB-PAAC_cryo and NB-SMs; nucleus stained with DAPI (blue), cytoskeleton stained with phalloidin (red). (f-g) Number of CTCs captured by the NB-PAAC_cryo from different volume of solutions at 1 and 10 CTC/mL (f), and their corresponding fluorescence microscopy images (g). (h) Total protein expression levels of the CTCs captured by the NB-PAAC_cryo compared with those of the control group. (i) Pearson correlation coefficient of proteomic profiles between the CTCs captured by the NB-PAAC_cryo and the control group. (j) Up-/down-regulation of differentially expressed proteins in the CTCs captured by the NB-PAAC_cryo compared to the control group. Panels (c), (d) and (f): data are presented as mean ± SD, n = 3; *p < 0.05, **p < 0.01, ***p < 0.001 and ns: not significant. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

After the successful preparation of the NB-PAAC_cryo cryogel, we investigated its CTC capture efficiency in dynamic fluid condition, using a closed-loop circulation set-up shown in Fig. 2b. Human epidermal carcinoma A431 cells, Human cervical cancer HeLa cells, human embryonic kidney 293 cells (HEK293), and human promyelocytic leukemia 60 (HL60) cells, with gradient EGFR expression levels (Fig. 2c and S22), were selected as CTCs or normal cell models in the dynamic capture experiments in aqueous solution. As shown in Fig. 2c, the NB-PAAC_cryo exhibited high CTC capture efficiency and specificity, which were positively correlated with cellular EGFR expression. At 1 × 10⁴ cell/mL and a flow rate of 4 mL/min, the capture efficiency for A431 cells (high EGFR expression) reached up to 91.1 ± 2.0% compared to 36.4 ± 1.6% for Hela cells (low EGFR expression), and 15.2 ± 2.7% and 4.7 ± 3.4% for EGFR-negative HEK293 and HL60 cells, respectively. In contrast, the PAAC_cryo without NB_EGFR demonstrated capture efficiencies below 20% for all the tested cell lines, further confirming that the capture was specific to EGFR-positive cells.

Besides, the interconnected macropores in the NB-PAAC_cryo (pore size: 50∼150 μm) were size-compatible with the normal blood cells (2∼20 μm) and CTCs (∼20 μm) to allow their free passage through the cryogel, while significantly increasing the collision frequency and contact time between these cells and the material surface through the structure-induced cell deceleration, thus enhancing the CTC capture efficiency. As shown in Fig. 2d and e, compared with NB_EGFR functionalized sepharose microspheres (NB-SMs) without the macroporous structure, the NB-PAAC_cryo exhibited markedly higher CTC capture efficiency in the hemodynamic simulation system across the cell concentrations range from 1 × 10⁴ down to 10 cell/mL. Especially, at low CTC concentration of 10 cell/mL, the capture efficiencies of the NB-PAAC_cryo reached 73.4 ± 2.3%. In contrast, no cells were captured by the NB-SMs. Further study demonstrated linear escalation in captured CTC number with increasing sample volume (Fig. 2f and g, S23 and S24), and the binding capacity of the 0.1 cm³ cryogel was measured to be ∼3.7 × 10⁵ cells. At an extreme low CTC concentration of 1 cell/mL, up to 2.7 × 10³ CTCs could be captured and enriched when 4000 mL sample was processed (approximately the total blood volume of an adult) using 0.1 cm³ of the NB-PAAC_cryo, exhibiting superior performance compared with existing technologies (typically less than 100 CTCs [57]).

Live/dead staining and colony-formation assay demonstrated that the captured CTCs retained their viability and proliferative capacity (Figs. S25 and S26). Proteomic analysis further confirmed that the key protein expression profiles of the cells were well preserved after the capture. As shown in Fig. 2h and i, among the 7053 proteins identified in the cell groups before and after capture by the NB-PAAC_cryo, total protein profiles showed no significant differences, and Pearson correlation analysis indicated a strong linear relationship between the two groups (R² = 0.99). Volcano plot analysis revealed that no proteins exhibited significant changes (p > 0.05, Fig. 2j). Heatmap clustering analysis further confirmed the high consistency of protein expression profiles (Fig. S27). Collectively, these results demonstrated that our NB-PAAC_cryo not only provides efficient and specific CTC capture but also preserved their viability biological characteristics for downstream analysis.

3.3. In vitro CTC capture of the NB-PAAC_cryo in simulated bloodstream

A further step was taken to assess the CTC capture performance of the NB-PAAC_cryo in whole blood. Such a task is much more challenging since blood is a complex biosample containing billions of blood cells, plasma proteins, and other biomolecules, which will greatly interfere with CTC capture.

Hemocompatibility of the NB-PAAC_cryo was evaluated prior to the CTC capture experiments in blood, since it is the primary requirement for blood-contacting materials. As shown in Fig. 3a, the hemolysis rates of the NB-PAAC_cryo cryogel were less than 2% across the tested concentrations (2-10 mg/mL), which were far below the safety threshold of 5% [58]. Besides, no significant impact on the blood clotting times (including blood clotting index, activated partial thromboplastin time, and prothrombin time) and complement (C3a/C5a) activation were observed when the blood samples were incubated with the NB-PAAC_cryo (Figs. S28 and S29). In addition, after five cycles of whole-blood perfusion through the NB-PAAC_cryo cryogel, no significant changes were observed in the count of blood cells or their size, as well as the plasma protein concentrations (Fig. 3b and S30). These results demonstrated the excellent hemocompatibility of our NB-PAAC_cryo.

Fig. 3.

In vitro CTC capture of the NB-PAAC_cryo in simulated bloodstream. (a) Hemolysis ratio and the corresponding photographs of the NB-PAAC_cryo treated erythrocyte suspensions in normal saline; normal saline, and deionized water suspended erythrocytes were used as the negative and positive controls, respectively, noted as Ctrl (−) and Ctrl (+). (b) Changes in complete blood count parameters after perfusion through the NB-PAAC_cryo, relative to normal control blood. (c) In vitro CTC capture efficiency of the NB-PAAC_cryo in a simulated bloodstream at different CTC concentrations. (d) Immunofluorescence microscopy images of CTCs captured on the NB-PAAC_cryo. Cells were stained with DAPI (blue), EGFR (green), and CD45 (red). Panels (a-c): data are presented as mean ± SD, n = 3; *p < 0.05, **p < 0.01, ***p < 0.001 and ns: not significant. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

After that we evaluated its CTC capture performance in whole blood under flow condition that mimicing the bloodstream. As shown in Fig. 3c, the NB-PAAC_cryo maintained excellent CTC capture efficiency in the complex whole blood environment at CTC concentrations ranging from 1 × 10³ down to 10 cell/mL. Notably, a capture efficiency of 72.3 ± 4.5% was achieved even at a low CTC concentration of 10 cell/mL. The high CTC capture efficiency of the NB-PAAC_cryo was attributed to its interconnected macroporous structure, which allowed the free passage of blood cells while significantly enhancing CTC-nanobody interactions through structure-induced cell deceleration. The cryogel exhibited satisfactory structural and functional stability under the circulatory condition. After 2 h of circulation at flow rates ranging from 5 to 20 mL/min, the NB-PAAC_cryo retained its interconnected macroporous structure, with no observable pore collapse or blockage (Fig. S31, Video S1). Furthermore, when the cryogel was pre-perfusion with healthy whole blood (without CTCs) at 5 mL/min for 2 h, no significant decrease in CTC capture efficiency was observed compared with the control without the blood-perfusion pretreatment (Fig. S32), indicating that the dynamic blood environment would not lead to antibody shedding, degradation, or masking by plasma proteins and blood cells.

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103689

The following are the Supplementary data related to this article:

What made this work more exciting was that the captured CTCs could be directly subjected to pathological examination, which is the clinical gold standard for tumor diagnosis. The NB-PAAC_cryo containing enriched CTCs could be directly processed for paraffin embedding, without any pretreatment of CTCs that frequently leads to loss of tumor information, and the obtained sections (Fig. S33) were fully compatible with standard clinical pathology workflows for cancer diagnosis. As shown in Fig. 3d and S34, for the control blood samples without the capture, the images were full of normal blood cells and no CTCs could be detected even at 1 × 10³ CTC/mL. When the CTC-containing blood samples passed through the NB-PAAC_cryo, CTCs, indicated by the high EGFR expression, were clearly observed in the immunofluorescence microscopy images of the cryogel pathological sections, with a CTC capture purity of ∼45%(Fig. S35a), even in samples with extremely low CTC concentrations of 10 cell/mL. The robust CTC capture efficiency and its compatibility with the standard clinical pathology workflows would contribute to the translational potential of our CTC capture strategy in cancer diagnosis.

3.4. In vitro CTC capture of the MNB-PAAC_cryo in simulated bloodstream

To achieve pan-cancer CTC capture, a cocktail of three nanobodies (NB_EGFR, NB_EpCAM and NB_VIM), each targeting a canonical tumor biomarker (EGFR, EpCAM and Vimentin, Fig. S36), was immobilized onto the PAAC_cryo cryogel, obtaining the multi-nanobody functionalized cryogel of MNB-PAAC_cryo (in vitro CTCs capture efficiency > 80%, Fig. S37). It demonstrated the highest and most robust CTC capture efficiency across different CTC subpopulations, with in vitro CTC capture efficiencies of 91.7% for A431, 87.2% for MCF7, and 80.8% for A549 from whole blood, outperforming all single- and pairwise-nanobody modified samples in terms of the broad-spectrum capture capability (Fig. S38). These results demonstrated the complementary effect of the three nanobodies on the capture capability against heterogeneous CTCs. Besides, it could also be observed that no competitive effect occurred in the binding activity of any single type of antibody, as the three-nanobody modified cryogel exhibited similar CTC capture efficiency when compared with the corresponding single nanobody modified cryogels. Further studies demonstrated the repeatability and stability of the MNB-PAAC_cryo. The coefficient of variation (CV) of replicate batches of the MNB-PAAC_cryo was 2.3%. In addition, after 20-day storage at 4 °C, the MNB-PAAC_cryo showed negligible changes in microstructure and capture efficiency when compared with the freshly-prepared one (Fig. S39).

3.5. In vivo CTC capture from the bloodstream via extracorporeal circulation

As shown in Fig. 4a, for the in vivo CTC capture, an extracorporeal circulation system incorporating the MNB-PAAC_cryo column (volume: 0.1 cm³) was established in subcutaneous tumor-bearing rat models with no detectable metastatic lesions. This strategy enabled the entire circulating blood volume of individuals with suspected cancer to interact with the MNB-PAAC_cryo cryogel for CTC capture. Besides, normal blood components such as blood cells and proteins can be returned to the body, avoiding the safety risks associated with large-volume blood loss (biosafety assessments demonstrated the excellent biocompatibility and biosafety profile of the MNB-PAAC_cryo based in vivo CTC capture strategy; Fig. S40).

Fig. 4.

In vivo CTC capture from the bloodstream via extracorporeal circulation in tumor-bearing rat models. (a) Photograph of the extracorporeal circulation device integrated with a MNB-PAAC_cryo column for CTCs capture from the bloodstream. (b-d) Immunofluorescence microscopy images of the MNB-PAAC_cryo after CTCs capture across (b) skin cancer, (c) lung cancer, and (d) liver cancer. The MNB-PAAC_cryo sections of skin cancer, lung cancer and liver cancer were stained with DAPI (blue), and CK18 (green). (e) Representative immunofluorescence images comparing captured CTCs and matched primary-tumor-tissue sections derived from skin-tumor-bearing rats; sections were stained for CK18, Ki67, CK7, VIM, EGFR and p40 (green), with nuclei counter-stained by DAPI (blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

After CTC capture via the extracorporeal circulation for 2 h, the MNB-PAAC_cryo cryogel was subjected to paraffin sectioning for pathological examination. As shown in Fig. 4b-d and S41, compared with the single-nanobody modified cryogel, the MNB-PAAC_cryo cryogel exhibited a broad-spectrum capture of diverse CTC subtypes across a wide range of cancer types, including skin cancer, lung cancer, and liver cancer. Over 3000 CTCs could be captured per rat (1 × 10⁴ cells/kg body weight) using only 0.1 cm³ of the MNB-PAAC_cryo, with a CTC capture purity ∼35% (Fig. S35b). In contrast, for the bare PAAC_cryo without any nanobody modification, negligible number of cells could be captured for all three cancer models (<10 cells by the 0.1 cm³ cryogel), and the only observed cells were indicated to be normal blood cells, as no signal of cancer biomarkers were observed in the immunofluorescence microscopy images (Fig. S42). Similar phenomenon was also observed for the healthy rat groups processed through MNB-PAAC_cryo (<10 cells captured). These results further demonstrated the specificity of the MNB-PAAC_cryo against CTCs.

Results of the Calcein-AM/PI staining demonstrated that captured CTCs are viable (Fig. S43). Moreover, the CTCs exhibited immunophenotypes consistent with primary tumor tissues for cancer diagnosis. In a double-blinded cancer identification experiment involving three different cancer types, we successfully identified cutaneous squamous cell carcinoma from lung adenocarcinoma and hepatocellular carcinoma by detecting a panel of characteristic antigens. As shown in Fig. 4e and S44, according to the immunofluorescence microscopy analysis of the cells captured by our strategy from the bloodstream, the positive expression of CK18 and Ki67 indicated that the captured cells were epithelial-origin cells with malignant proliferation [59], suggesting a tumor-derived phenotype. The negligible expression of CK7 (glandular tumor biomarker [60]) and vimentin (VIM, mesenchymal tumor biomarker [61]) exclude the possibility of lung adenocarcinoma and hepatocellular carcinoma. Furthermore, the robust EGFR and p40 expression substantiated the cutaneous lineage and squamous differentiation of the captured cell [62,63], providing strong evidence to confirm the diagnosis of cutaneous squamous cell carcinoma. These diagnostic results were highly consistent with the biopsy findings of the primary tumor tissues, demonstrating the great potential of our strategy for CTC-based liquid biopsy in cancer diagnosis. Table 1 presents a multi-parameter comparison of our MNB-PAAC_cryo cryogel with other representative in vivo CTC capture platforms.

Table 1.

Comparison of our MNB-PAAC_cryo cryogel with other representative in vivo CTC capture platforms.

Parameter	MNB-PAACcryo (this work)	MagWIRE [43]	CellCollector [40]	CTC-iChip [64]
animal model	rat	pig	human	human
cancer type	skin, lung, and liver cancers	H1650 cell line injected into healthy pig	breast and lung cancers	prostate, breast, lung, pancreas, and melanoma cancers
operating mode	extracorporeal circulation	insert into auricular vein	insert into cubital vein	extracorporeal circulation
material volume	0.1 cm3	0.027 cm3	0.01 cm3	not mentioned
operating time	120 min	60 min	30 min	> 120 min
capture efficiency	in vitro: >80% in vivo: n/a	in vitro: 49 ± 8% in vivo: 1-8%	in vivo: ∼0.0016%	n/a
captured number	>3000 CTCs	not mentioned	16 CTCs	53 CTCs
pathological examination compatibility	yes	no	no	no

4. Conclusions

In summary, this study proposes a robust strategy toward high-quantity enrichment of CTCs from the bloodstream via a multinanobody-functionalized carboxymethyl cellulose cryogel of MNB-PAAC_cryo. By addressing the trade-off between pore size and mechanical property, the MNB-PAAC_cryo cryogel sets up a sieving barrier in the bloodstream via structure-induced cell deceleration. It allows free passage of blood cells through its interconnected macropores, while significantly enhances CTC-nanobody interactions at the internal pore surface, contributing to a high CTC capture efficiency of 72.3 ± 4.5% in the bloodstream at 10 CTC/mL. Consequently, up to 2.7 × 10³ CTCs could be harvested from 4000 mL blood at 1 CTC/mL using 0.1 cm³ of the MNB-PAAC_cryo cryogel. In vivo experiments on cancer rat models further demonstrated that the MNB-PAAC_cryo could capture diverse CTC subtypes across multiple cancers, and the captured CTCs exhibited consistent information with the primary tumor tissues in the pathological examination. With continued refinement in device engineering and perfusion protocols for clinical applications, our strategy shows great potential to advance CTCs from dispensable prognostic biomarkers towards standardizable diagnostic tools in clinical oncology. Besides, it could also contribute to a more personalized cancer therapy, as the high-quantity enrichment of CTCs would greatly improve the success rate and fidelity of tumor organoid establishment for subsequent drug screening. Making full use of our strategy will accelerate the era of personalized medicine in cancer care.

CRediT authorship contribution statement

Liang Tai: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Hanting Wang: Writing – original draft, Visualization, Validation, Methodology, Formal analysis, Data curation. Yongjie Wang: Validation, Methodology, Investigation, Data curation. Fen Zeng: Validation, Methodology, Investigation, Data curation. Zhen He: Methodology, Investigation, Data curation. Jiewen Chen: Validation, Methodology, Investigation. Zhenqiang Shi: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization. Lingyun Jia: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lingyun Jia, Zhenqiang Shi has patent pending to Dalian University of Technology. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.