High-temporal-resolution on-site multiplex biomarker monitoring in small animals using microfluidic digital ELISA

Abstract

Time-course monitoring of blood biomarkers with rapid turnaround has the potential to revolutionize the diagnosis, stratification of phenotypes, and therapeutic/prognostic approaches for various acute inflammatory diseases in both clinical and preclinical studies. Current approaches, however, are hampered by slow turnaround times and large sample volume requirements, limiting the exploration of disease mechanisms and therapeutic strategies. Here, we developed a microfluidic digital ELISA platform prototype, combining single-molecule counting with whole blood assay capability for the first time from small animal models. This platform is semi-automated and enables repeated, rapid biomarker monitoring with just 3.5 μL of whole blood collected from the tail. Our platform demonstrated high sensitivity and multiplexity, allowing real-time cytokine profiling within a 2-h turnaround. Using a murine sepsis model, we achieved precise temporal monitoring of cytokine levels, demonstrating prognostic capability by correlating early-stage cytokine levels with a liver-injury biomarker. This microfluidic platform enables high temporal resolution and rapid monitoring of biomarker dynamics in a single mouse using freshly collected whole blood, significantly reducing the number of animals needed for preclinical studies. This technology has strong potential to transform ICU therapeutic strategies and preclinical research, enabling personalized treatment based on real-time biomarker profiles.

Keywords: Digital ELISA; Microfluidics; Sepsis; Time-course; Whole blood.

Figure 1.. Concept of PEdELISA-enabled time-course mouse model studies

(A) Schematic of conventional time-course mouse model study requiring animal sacrifice and serum collection at a single time point, where each time point contains 3 infected and 1 sham mouse. This method is retrospective, with daily time resolution, large sample volume requirements, and high costs due to the number of animals needed. (B) Schematic of the new mouse model study enabled by PEdELISA, which eliminates the need for animal sacrifice at each time point. This approach allows real-time, prospective monitoring with hourly time resolution using tail whole blood collection (3.5μL), significantly reducing costs and the number of animals required, and enabling the collection of within-subject data over time.

Figure 2.. Engineering Prototype of PEdELISA

(A) Chip design concept and image of the PEdELISA chip, illustrating the pre-equilibrium concept and pre-patterned microarray technology for multiplex biomarker detection. Scale bar: 20 μm. (B) Current workflow of the PEdELISA assay, which includes whole blood incubation, detection antibody labeling, streptavidin-HRP labeling, substrate loading, oil sealing, and fluorescence imaging. (C) Design and real image of the PEdELISA fluidic system (inset: manifold with whole blood incubation). (D) PEdELISA reader system and graphic user interface (GUI) for high-throughput, automated fluorescence imaging. (E) Automated data analysis powered by AI. Fluorescence and brightfield images obtained from the PEdELISA assay were analyzed by a pre-trained convolutional neural network (CNN).

Figure 3.. Whole Blood Assay Characterization

(A) 4-plex PEdELISA titration standard curves from 0.32 pg/mL to 5000 pg/mL in ELISA dilution buffer (1% bovine serum albumin). Each data point represents the average signal from three independent chips. The digital immunoassay signal (average enzyme per bead, AEB) was fitted with four-parameter logistic (4PL) curves. The dotted line represents the signal level from a blank solution plus 3 times the standard deviation (3σ) for each cytokine, which is used to estimate the limit of detection (LOD). (B) Assay specificity test with “all-spike-in,” “single-spike-in,” and “blank” (negative) samples of recombinant cytokine marker(s) at 400 pg/mL in ELISA dilution buffer. (C) Correlation between multiplex PEdELISA and conventional single-plex ELISA results using diluted mouse plasma from tail blood collection. Pearson’s R² values were calculated for MCP-1 (R²=0.849), CXCL-1 (R²=0.923), CCL-11 (R²=0.981) and IL-6 (R²=0.957). (D) On-chip blood culture test at 5 minutes, 1 hour, 3 hour and after wash, validating chip surface passivation. Mouse whole blood was diluted 10-fold in 1x PBS. (E) Correlation between whole blood and plasma using Heparin as the anticoagulant measured by PEdELISA using freshly collected CS mouse samples. Pearson’s R² values were calculated for MCP-1 (R²=0.925), CXCL-1 (R²=0.877), CCL-11 (R²=0.950) and IL-6 (R²=0.977). (F) Correlation between whole blood and plasma using EDTA as the anticoagulant measured by PEdELISA using freshly collected CS mouse samples. Pearson’s R² values were calculated for MCP-1 (R²=0.859), CXCL-1 (R²=0.930), CCL-11 (R²=0.965) and IL-6 (R²=0.978).

Figure 4.. PEdELISA high-temporal-resolution whole blood biomarker detection

(A) High-resolution retrospective experiment to evaluate CS-induced sepsis severity, disease development dynamics, and response to antibiotics at 6 hours post-injection. Whole blood samples were collected every hour for the high dose case (18uL/g) and every 2 hours for the low dose (14uL/g) and control (saline) cases for a total of 12 hours. Rectal (body) temperature was monitored every hour. Antibiotics (ceftriaxone and metronidazole) were administered at the 6-hour after infection. (B) Photographs of the animal and equipment setup. (i) A cage housing the test mice was placed next to the manifold operation unit for the PEdELISA chip. (ii) Whole blood was drawn from the tail vein of one of the mice. (iii) The collected blood sample was diluted by a buffer solution within the PCR tube and subsequently loaded into the inlets of the manifold for PEdELISA biomarker analysis on the microfluidic chip. (iv) The rectal temperature of the mouse was measured. (C) Clinical scores recorded for the mice based on their response to a finger poke, signs of encephalopathy, and overall appearance, evaluated every 2 hours. (D) Real-time 4-plex whole blood detection results of 10 mice in parallel using PEdELISA, showing cytokine levels and body temperature trends over time. 3.5uL of whole blood samples were collected from the tail vein every 2 hours at 0, 2, 4, 6, 8, and 24 hours, diluted, and assayed immediately within 2 hours, showcasing the prospective diagnostic detection capability. Antibiotics were administered at the 6-hour after infection. (E) Correlation between the 4 cytokines, temperature, and clinical scores of the 10 mice at all time points. Pearson’s R² values were calculated and used to determine the correlation quality.

Figure 5.. Association Between Cytokine Levels and Liver Injury

(A) Correlation between IL-6 levels at 0, 6, 8, 24 hours post-infection and liver injury marker Aspartate Aminotransferase (AST) at 24 hours. Cytokine levels measured at 8 hours post infection had the strongest association with liver injury 24 hours post infection. (B) F-statistic values of all cytokines at different times post-infection, correlated with the liver injury marker AST at 24 hours, as measured with the PEdELISA platform, illustrating the importance of time lags in the relationship between immunopathology and organ injury. The dotted line indicates the F(1,23) value corresponding to p = 0.05. n=24 mice.