Phenol Red Thread-based Sampling Procedure for Untargeted Tear Fluid Lipidomics in Biomarker Discovery

Tear film is the outermost barrier that protects the ocular surface from pathogens and maintains ocular homeostasis^1,2, while the lipid layer in tear film is the key component for preventing tear evaporation and maintaining tear film stability^3,4. Dysregulation of the tear lipid composition has been linked to various eye diseases, including dry eye disease (DED), meibomian gland dysfunction (MGD), and allergic conjunctivitis^5,6,7,8. As a result, detailed characterization of tear lipidomics has become a critical and trending research direction for a more comprehensive understanding of the molecular pathologies underlying these ocular conditions⁹.

While the tear lipidomic approach holds significant clinical importance, a critical challenge in tear lipidomics is establishing a standardized, reproducible collection and analysis workflow. Current tear sampling methods, Schirmer's strips and microcapillary tube, each present distinct limitations. Schirmer's strips have a large contact area with the ocular surface that often causes ocular irritation and reflex tearing^9,10. In comparison, microcapillary tube collection is highly dependent on operator technique and can yield inconsistent results when performed by different personnel¹¹. Additionally, improper use of these sampling tools may also carry a risk of ocular surface injury. These limitations could compromise the accuracy and reproducibility of downstream data^12,13. Furthermore, tear collection methods that require complex handling procedures may limit their accessibility and standardization across different laboratories and clinical settings. Therefore, there is a pressing need to establish a non-invasive, robust, and reproducible tear collection and processing workflow that can preserve the native tear lipid profile.

Recent studies have highlighted the utility of phenol red thread (PRT) for downstream LC-MS/MS workflows analyses in proteomic and metabolomic applications^14,15. PRT offers inherent advantages as a tear collection tool. Its thin structure, minimally invasive nature, and straightforward sampling procedure could minimize technical complexity and reduce operator-dependent variability, while being well-tolerated by most subjects^14,16. However, a standardized workflow specifically for PRT-based tear lipidomic analysis has not yet been established.

This protocol introduces an integrated workflow for reproducible tear lipidomics: a user-friendly PRT-based tear collection method adapted for lipid analysis, an optimized methyl tert-butyl ether (MTBE)/methanol lipid extraction protocol with defined solvent-to-sample ratios and storage controls^17,18, and a modified lipid identification method using high-resolution LC-MS/MS. This standardized approach enables a consistent tear lipidomic profiling for downstream biomarker discovery and disease characterization. Also, it is particularly well-suited for subjects with low tear volume or reduced tolerance for foreign body sensation (Figure 1).

Figure 1: Schematic overview of the Phenol Red Thread (PRT)-based workflow for human tear lipidomics. Tear fluid is collected from subjects utilizing the PRT approach, which enables consistent sample acquisition with minimal discomfort. Sample processing involves sequential MTBE/methanol biphasic extraction, SpeedVac-mediated concentration, and reconstitution of tear lipid extracts for liquid chromatography-tandem mass spectrometry (LC-MS/MS). Lipidomic profiles are subsequently acquired via LC-MS/MS and subjected to annotation and identification using LipidSearch software. Please click here to view a larger version of this figure.

The study was approved by the Institutional Review Board (IRB) of The Hong Kong Polytechnic University. The subjects provided written informed consent before participation in the study. The reagents and the equipment used are listed in the Table of Materials.

1. Phenol Red Thread (PRT)-based collection of human tear fluid

1. Wear gloves and disinfect the workstation to prevent sample contamination.
2. Make sure the PRT package is sealed and not expired before opening it.
3. Hold the thread by the end opposite the bent hook. Always avoid touching the bent hook half of the PRT.
4. Ask the subject to look at the superior nasal direction.
5. Pull down the subject's lower eyelid gently and insert the PRT with the bent hook positioned in the lower temporal palpebral conjunctiva near the outer canthus of the subject's eye (Figure 2).

Figure 2: Position of the phenol red thread during tear collection. Please click here to view a larger version of this figure.

6. Ask the subject to close their eyes gently and tilt his/her head slightly forward to prevent the PRT from touching his/her facial skin.
   NOTE: Tear absorbed would gradually induce a red dye front, indicating the sampled region of PRT due to the pH-sensitive phenol red. To ensure enough tear fluid for lipidomic analysis, collecting tear volume equivalent to 50 mm of PRT is preferred, typically completed within 2 min.
7. Pull down the subject's lower eyelid gently and pull out the sampled PRT. Avoid touching the sampled region.
8. Record the length of the sampled region of the PRT and transfer the PRT into a new sample tube with the aid of sterile forceps.
   NOTE: If the sampled PRT length is <50 mm, kindly ask the subject to rest for 1-3 min and repeat the collection with the same eye. Transfer each sampled PRT to a separate sample tube.
9. Label and cap the sample tubes properly.
10. Clean the gloves and forceps thoroughly with methanol and clean paper wipes to remove any tear residuals or contaminants, and then let them dry.
11. Repeat steps 1.2-1.10 for the other eye.
12. Seal the sample tubes with parafilm and immediately store the sample at -20°C for temporary storage or -80°C for long-term storage. Avoid unnecessary freeze-and-thaw cycles and protect the sampled PRT from light.

2. Processing of sampled PRT

1. Pre-wash the micro-scissors with methanol before processing the PRTs.
2. Discard the bent hook end (i.e., first 3 mm) of the PRTs (Figure 3).

Figure 3: Physical appearance of the phenol red thread (PRT). (A) Unused PRT with bent hook end. (B) Used PRT with the indicated section cut and collected for lipid extraction. Please click here to view a larger version of this figure.

3. Cut the sampled region of the PRT (i.e., region above the phenol red dye front) into 2 mm pieces and transfer these pieces to a clean 1.5 mL organic solvent-tolerant microcentrifuge tube.

3. MTBE/methanol biphasic separation for lipid extraction from sampled PRT

CAUTION: MTBE and methanol are flammable and volatile. Use in a well-ventilated area or fume hood. Avoid open flames and sparks. Wear gloves and a lab coat. Wash hands after handling.

1. Place the samples on ice.
2. Add 232 µL of ice-cold methanol (MS grade) to each sample and vortex for 15 s.
3. Sonicate the samples in a pre-cooled ultrasonic cleanser for 15 min.
4. Add 774 µL of MTBE (HPLC grade) and 194 µL of deionized water to each sample and vortex for 15 s. The resultant volume ratio of MTBE:methanol:water in the sample should be 4:1.2:1 (v/v/v).
5. Sonicate the samples again in a pre-cooled ultrasonic cleanser for 15 min.
6. Incubate the sample in the thermomixer for 8 h at 4 °C, with shaking at 1,200 rpm.
7. Let the mixture sit at room temperature for 10 min.
8. Centrifuge the resultant mixtures at 10,000 × g for 10 min at 4 °C.
9. After centrifugation, the mixtures are separated into two phases (Figure 4).
   NOTE: Upon completion of phase separation, inspect the sample for two distinct layers: the upper organic phase (MTBE-rich) should appear transparent with a slight milky opacity, while the lower aqueous phase exhibits a yellow coloration attributable to the presence of phenol red indicator. Proceed only if these phase characteristics are observed, as shown in Figure 4.

Figure 4: Phase separation of lipid extraction using MTBE-methanol-water: upper organic phase and lower aqueous phase. Please click here to view a larger version of this figure.

10. Collect 700 µL of the upper organic fraction containing lipids carefully and transfer to a pre-chilled new 1.5 mL microcentrifuge tube.
11. Discard the lower aqueous phase containing residual MTBE and methanol as hazardous chemical waste. Collect waste in labeled, sealed containers and dispose of it according to institutional hazardous waste protocols and local regulations. Do not pour solvents down the drain.
12. Dry the lipid extracts using a refrigerated SpeedVac concentrator for 4 h at 4 °C.
    NOTE: To ensure complete drying, periodically check the sample dryness starting at 2 h. Extend the drying time as necessary until samples are fully dried.
13. Store the dried lipid extracts at -20 °C for short-term storage or -80 °C for long-term storage.

4. Reconstitution of extracted tear lipid for LC- MS/MS analysis

1. Reconstitute the dried lipid extracts with 50 µL of the ice-cold methanol (MS grade)/chloroform (HPLC grade) mix (1:1, v/v) as the solvent (1 µL of solvent per 1 mm of PRT used) and follow by sonication in a pre-cooled ultrasonic cleanser for 15 min.
2. Centrifuge the reconstituted lipid extracts at 14,000 × g and 4 °C for 10 min.
3. Carefully transfer 40 µL of the supernatant to an autosampler glass vial with a glass insert installed.
4. Tightly cap the vial with a non-slit septum open top cap and ready for LC-MS/MS analysis.

5. Sample acquisition by LC-MS/MS

1. For Liquid Chromatography, set the column chamber and sampler temperature at 50 °C and 4 °C, respectively.
2. For each injection, load 5 µL of the sample onto a reverse-phase LC column (C18; 1.7 µm, 100 mm × 2.1 mm) and fractionate the lipids at a flow rate of 0.3 mL/min in a 24 min separation gradient.
   1. Use mobile phase A comprising a mixture of 60:40 acetonitrile (MS grade):water (v/v) with 5 mM ammonium formate (MS grade) and 0.1% (v/v) formic acid (MS grade) and mobile phase B containing 90:10 isopropanol (MS grade):acetonitrile (v/v) with 5 mM ammonium formate and 0.1% (v/v) formic acid.
   2. Use the following gradient: 0-2 min: 40% B; 2-2.5 min: 40% B; 2.5-3 min: 58% B; 3-18 min: 99% B; 18-20 min: 99% B; 20-20.1 min: 40% B; 20.1-24 min: 40% B.
3. For the Orbitrap Mass Spectrometer, set the spray voltage to +3.5 kV for positive mode or -2.5 kV for negative mode.
4. Adjust the sheath gas flow rate to 45 Arb and the auxiliary gas flow rate to 10 Arb.
5. Set the sweep gas flow rate to 2 Arb.
6. Set the ion transfer tube temperature to 300 °C and the vaporizer temperature to 320 °C.
7. Use data-dependent acquisition (DDA) mode.
8. For MS1, set the resolution to 120,000, scan range to 100-2000 m/z, standard AGC target, auto injection time, dynamic exclusion to 6 s, and intensity threshold at 5.0e4.
9. For MS2, set the resolution to 15,000, auto scan range, standard AGC target, dynamic injection time, isolation window at 1.6 m/z, collision energy type as higher-energy collisional dissociation (HCD), and normalized collision energy (NCE) to stepped values of 15, 30, and 40.
10. Analyze raw data with LipidSearch software or other compatible platforms.

Composite tear samples were collected from 16 healthy volunteers at three independent visits spaced two weeks apart (R1, oldest batch; R2, R3 collected sequentially two weeks after each). Using LipidSearch with stringent criteria (i.e., Signal-to-Noise Ratio ≥100, grade C or above, and ion intensity ≥30,000), we identified 773, 890, and 1,025 unique lipid species in R1-R3, respectively (Figure 5A). Positive ion mode identified 1,302 lipid species with 26.0% Grade A (i.e., both lipid class and all fatty acid chains belonging to a given lipid were completely identified; 338 lipids), 13.2% Grade B (i.e., full identification of lipid class and partial identification of fatty acid chains), and 60.8% Grade C (i.e., lipid class specific ion or fatty-acid-derived product ions were detected). Negative ion mode identified 257 unique lipid species with 8.2% Grade A (21 lipids), 22.9% Grade B, and 68.9% Grade C (Figure 5B). Combined analysis detected 1,559 total unique lipid species across both modes. Despite storage-related variations, 394 species were consistently identified across all three sample sets (Figure 6). Intra-subject coefficient of variation across the three visits was 21.5% (positive mode) and 31.3% (negative mode), with 23 lipid classes recovered across both ionization modes (Figure 7A,B).

Compared to prior tear lipidome studies4,19, this workflow detected 1,559 unique species with substantially improved structural characterization. The expanded coverage reflects high-resolution LC-MS/MS separation, stringent identification criteria minimizing artifacts, and complementary dual-mode analysis. Notably, unlike prior studies lacking identification confidence metrics, this workflow provides transparent Grade-level reporting: positive ion mode yielded 338 Grade A identifications (highest structural confidence), while negative ion mode contributed complementary lipid diversity. The consistent core lipid set (394 species) across different storage durations, combined with robust positive mode Grade A representation (26%) and acceptable intra-subject CVs, demonstrates reliable detection suitable for quantitative tear lipidomic studies.

Figure 5: Lipid identification summary and quality assessment. (A) Total unique lipid species identified in each sample group (R1, R2, R3) by ionization mode. (B) Grade-level confidence distribution of lipid identifications in positive and negative ion modes. Please click here to view a larger version of this figure.

Figure 6: Venn diagram showing overlapping unique lipid species identified across three sample groups. Each circle represents the union of unique lipid species found in one sample group, while the intersections show lipid species that were commonly found in two or all three groups. This visualization highlights consistent and distinct lipid species among the groups. Please click here to view a larger version of this figure.

Figure 7: Consistent lipid class distribution across sample replicates. Distribution of unique lipid species identified within each lipid class by LC-MS/MS in (A) positive ion mode and (B) negative ion mode. Bars represent the number of unique lipids consistently identified across all three sample sets (R1-R3). Major lipid classes shown include: glycerolipids (TG, DG), sphingolipids (Cer, SM, Hex1Cer), phospholipids (PC, PE, PI, PS, PEt), lysophospholipids (LPC, LPE), and other minor species. Please click here to view a larger version of this figure.

Advantages of PRT-based tear sampling and lipid extraction
The PRT-based tear lipidomics workflow presented here provides a minimally invasive and practical sampling method for comprehensive tear lipid profiling. Compared to Schirmer strips, which require prolonged contact and larger sample volumes, PRT sampling is rapid, can be completed within 2 min, well-tolerated by diverse populations, including those with reduced tear volume or heightened ocular sensitivity, and yields tear samples compatible with high-resolution LC-MS lipidomics^14,20. Unlike microcapillary tube collection, which is time-consuming and requires precise technical handling and operator expertise, PRT-based sampling requires minimal training and can be readily performed in clinical settings with high reproducibility and ease of use¹⁴. Regarding the lipid extraction method, MTBE/methanol-based phase separation avoids chloroform toxicity while providing comparable or superior lipid recovery to traditional Folch and Bligh-Dyer methods for most lipid classes, though it may show slightly reduced recovery for certain polar lysophospholipids (LPC, LPE). This workflow successfully identified over 700 unique tear lipid species across 23 lipid classes, substantially expanding previous tear lipidome datasets^4,19.

Technical Considerations and troubleshooting
Several factors are critical for reproducible results. The reagent volumes specified in this protocol (232 µL methanol, 774 µL MTBE, 194 µL water for lipid extraction; 50 µL methanol/chloroform for reconstitution) are optimized for a standard 50 mm PRT sample, yielding an MTBE/methanol/water ratio of approximately 4:1.2:1 (v/v/v). This ratio was selected based on optimization studies demonstrating robust extraction efficiency and reproducibility across diverse sample matrices^18,21. Sample-to-solvent ratio is a critical parameter that directly influences lipid yield and analytical sensitivity. For samples with collected PRT lengths different from 50 mm, users must adjust solvent volumes proportionally according to the per-mm PRT-to-solvent ratios, where lipid extraction requires 24 µL MTBE-methanol-water mix per 1 mm of PRT, and reconstitution requires 1 µL chloroform-methanol mix per 1 mm.

Strict adherence to this solvent ratio is critical for optimal phase separation and signal intensity. If insufficient PRT is collected (e.g., <50 mm) without reducing reconstitution solvent volumes proportionally, the resulting low sample concentration yields diminished MS signal intensity, readily masked by background noise, substantially reducing the number of identifiable lipid species and spectral quality. The reduced solute concentration diminishes ionization efficiency and detection capacity in electrospray ionization-based MS workflows. Conversely, if PRT exceeds the standard amount (e.g., >50 mm) without proportionally increasing solvent volumes, excessive sample material absorbs and retains water, reducing the aqueous phase capacity to solubilize hydrophilic components. This inefficient phase separation causes the lower aqueous phase to turn red instead of the expected yellow, indicating incomplete partitioning. Phenol red and other hydrophilic contaminants then migrate into the upper organic phase. Upon drying, such samples display deep red or brownish yellow coloration, a visual indicator of phenol red contamination and ion suppression risk in subsequent MS analysis. If this coloration occurs, redissolve the sample and repeat the lipid extraction steps 3.2-3.12 to achieve clean phase separation and remove interfering substances. Careful documentation of PRT length for each sample is essential to maintain analytical consistency^22,23. A gradual loss of detectable lipid species was observed in older samples (R1). This highlighted the importance of immediate transfer of sampled PRT to a dark environment at -80 °C for preventing enzymatic and oxidative degradation^9,24,25. This also underscored the importance of standardized storage protocols for large cohort studies.

Limitations
This workflow has several notable limitations. The requirement for participants to maintain upward gaze during PRT insertion may be challenging for individuals with nystagmus or severe ocular surface discomfort. Additionally, the workflow has been optimized for high-resolution Orbitrap MS/MS platforms and may require additional validation for other MS platforms. Besides, previous studies have shown that the MTBE extraction method might exhibit reduced recovery for certain polar lipid classes, particularly lysophospholipids (LPC, LPE), compared to chloroform-based approaches such as Folch and Bligh-Dyer methods²⁶. Despite these limitations, the PRT-based approach remains a practical and minimally invasive alternative for tear biomarker discovery in ophthalmology and systemic disease research.

Future applications and clinical significance
The enhanced lipid coverage provided by this workflow supports multiple translational applications. In the near future, tear lipid profiling can identify and validate biomarkers for ocular diseases such as dry eye disease and meibomian gland dysfunction or other systemic conditions associated with tear lipid alterations²⁷. The rapid sampling and standardized processing enable longitudinal studies to monitor disease progression and treatment responses. Long-term perspectives include integration with proteomics and metabolomics for multi-omics profiling, which would enable personalized medicine approaches and point-of-care diagnostics^12,28,29. Given the minimally invasive nature and robust lipid identification capabilities, PRT-based tear lipidomics is well-positioned for large-scale population screening and precision ophthalmological applications.