Boost Immunopeptidomics Efficiency with Microfluidic Automation

Microfluidic automation for precision handling of sparse samples

AMF - Xiaokang Li - Unleashing New Frontiers in Immunopeptidomics - AMF 4

Despite the recent advances in mass spectrometry (MS) instrumentation and data analytics tools, MS-based immunopeptidomics is still limited by conventional sample preparation methods. The current techniques are frequently laborious, compromising assay sensitivity and hindering large-scale clinical applications. These methods also require bulky sample sources, while patient-derived clinical samples are often restricted.

Current technology can provide the tools to create integrated microfluidic automation systems that minimise handling and sample loss, maximise sample enrichment, and enhance assay sensitivity for clinical studies in immunopeptidomics. “A more sensitive immunoaffinity purification workflow that can enrich the immunopeptidome of needle-biopsy-sized tumour samples is urgent for clinical immunopeptidomics studies,” claim XiaokangLi and colleagues, who engineered an automated fluidic platform using advanced and easy-to-use microfluidics technology to address these challenges.

AMF - Xiaokang Li - Unleashing New Frontiers in Immunopeptidomics - AMF 3
They created a programmable microfluidic automation platform that uses two syringe pumps to load samples and supply all reagents into the immunoaffinity purification chip (chip-IP). One significant advantage of their design is that it directly integrates the C18 cartridges for sample cleanup through a programmable switch valve, bypassing all the manual handling of sample transfers and washing steps.

Due to its ultralow internal volume and accurate liquid handling, this advanced fluidics setup significantly reduces reagent consumption and sample volume to the sub-millilitre range.

Their results describe how an advanced automated fluidic workflow can improve peptide purification and enrichment from cell lines and small tissue samples, enhancing assay sensitivity. The study also illustrates how microfluidic automation technology is promising for identifying difficult-to-explore but clinically relevant molecules, such as non-canonical peptides and tumour-associated antigens.

Materials & Methods

Their results describe how an advanced automated fluidic workflow can improve peptide purification and enrichment from cell lines and small tissue samples, enhancing assay sensitivity. The study also illustrates how microfluidic automation technology is promising for identifying difficult-to-explore but clinically relevant molecules, such as non-canonical peptides and tumour-associated antigens.

Microfluidic chip

The chip contained ~250,000 micropillars in a 50-cm-long fluidic channel on a silicon substrate. For homogenous inter-pillar flow distribution, the authors arranged the pillars diagonally and added additional microstructures to assist the in-flow reaction.

The silicon micropillars were thermally treated to provide a chemically active silicon surface for the silane molecules bearing protein-binding groups as a protein coating strategy. Additionally, adding an intermediate layer of protein A/G optimised antibody orientation and protein binding before coating the antibodies. Reinforcing the antibody-protein A/G by a crosslinking reaction reduced antibody coelution and may enable the reusing of the chips.

The chip was thermally bonded to create a leak-proof hermetic channel and clamped to reinforce and enhance its mechanical robustness. These measures avoided leaking and sample loss during the experiments.

Microfluidic automation

The authors integrated the chip into an automated fluidic control system by coupling two programmable syringe pumps to the chip: the IP pump and the C18 desalting pump.

The IP pump handled the antibody coating/crosslinking and IP reaction, while the C18 desalting pump controlled the peptide cleanup steps using C18 materials.

The system automatically handled the sequential injection of multiple reagents at different flow rates during the assay’s various steps, bypassing the manual handling of several tedious washing processes common to IP experiments.

Each pump included a 12-port programmable distribution valve controlled by Python programming via serial connection. Python scripts operated all the experimental steps, managing the different flow rates required for antibody coating, chip-IP, and peptide cleanup, thus handling reagent volumes efficiently.

The peptide cleanup control system used C18 cartridges and was linked to the chip-IP device via an automated six-port switch electric rotary valve that:

  1. Automatically separated the IP (before the acid elution step) from the C18 cartridge conditioning
  2. Sent the IP elution straight into the C18 cartridge without extra handling

The designed microfluidic automation platform allowed the eluted peptides to flow directly into the preconditioned C18 cartridge while the IP reaction was ongoing, automatically streamlining the HLA immunopeptidome purification process.

Immunoaffinity purification and LC-MS/MS analyses

Frozen cell pellets and disrupted tumour tissues were lysed in an ice-cold lysis buffer. The lysates were cleared by centrifugation, and the supernatant was used for either the chip-IP or column-IP procedure.

The authors applied the lowest flow rate for the Chip-IP when loading the lysate supernatant into the antibody-coated chip to maximise immunoaffinity reactions. For the Column-IP, they followed pre-established and published protocols.

All samples were analysed in an LC-MS/MS system composed of an Easy-nLC 1200 HPLC system coupled to a Q Exactive HF-X mass spectrometer (ThermoFisher Scientific).

Data analyses

The authors adopted the data-independent acquisition (DIA) and the data-dependent acquisition (DDA) for analysing the resulting peptides. Library-directed DIA has been a widely used method for proteome and immunopeptidome analyses.

For the experiments using human B-cells, they built an RA957-specific spectral library in Spectronaut containing the DIA and DDA data from both the chip-IP and column-IP methods and used it to compare the results from both techniques.

However, due to the scarcity of human-derived samples, generating a spectral library based on DDA and DIA data from replicates is challenging. Thus, the authors created a library-independent “direct DIA”, a technique not yet applied to immunopeptidomics.

For non-canonical peptide identification, the authors combined information from Ouspenkaia et al. database on proteins translated from novel or unannotated open reading frames (nuORF) and the DIA data from their tissue samples.

Finally, the researchers created a generic spectral library, the “Lausanne Lib,” by combining the available raw MS files from previous samples measured in their laboratory with the UniProt and the nuORF database.

Microfluidic chip

Microfluidic chip

The chip contained ~250,000 micropillars in a 50-cm-long fluidic channel on a silicon substrate. For homogenous inter-pillar flow distribution, the authors arranged the pillars diagonally and added additional microstructures to assist the in-flow reaction.

The silicon micropillars were thermally treated to provide a chemically active silicon surface for the silane molecules bearing protein-binding groups as a protein coating strategy. Additionally, adding an intermediate layer of protein A/G optimised antibody orientation and protein binding before coating the antibodies. Reinforcing the antibody-protein A/G by a crosslinking reaction reduced antibody coelution and may enable the reusing of the chips.

The chip was thermally bonded to create a leak-proof hermetic channel and clamped to reinforce and enhance its mechanical robustness. These measures avoided leaking and sample loss during the experiments.

Microfluidic automation

Microfluidic automation

The authors integrated the chip into an automated fluidic control system by coupling two programmable syringe pumps to the chip: the IP pump and the C18 desalting pump.

The IP pump handled the antibody coating/crosslinking and IP reaction, while the C18 desalting pump controlled the peptide cleanup steps using C18 materials.

The system automatically handled the sequential injection of multiple reagents at different flow rates during the assay’s various steps, bypassing the manual handling of several tedious washing processes common to IP experiments.

Each pump included a 12-port programmable distribution valve controlled by Python programming via serial connection. Python scripts operated all the experimental steps, managing the different flow rates required for antibody coating, chip-IP, and peptide cleanup, thus handling reagent volumes efficiently.

The peptide cleanup control system used C18 cartridges and was linked to the chip-IP device via an automated six-port switch electric rotary valve that:

  1. Automatically separated the IP (before the acid elution step) from the C18 cartridge conditioning
  2. Sent the IP elution straight into the C18 cartridge without extra handling

The designed microfluidic automation platform allowed the eluted peptides to flow directly into the preconditioned C18 cartridge while the IP reaction was ongoing, automatically streamlining the HLA immunopeptidome purification process.

Immunoaffinity purification
& LC-MS/MS analyses

Immunoaffinity purification and LC-MS/MS analyses

Frozen cell pellets and disrupted tumour tissues were lysed in an ice-cold lysis buffer. The lysates were cleared by centrifugation, and the supernatant was used for either the chip-IP or column-IP procedure.

The authors applied the lowest flow rate for the Chip-IP when loading the lysate supernatant into the antibody-coated chip to maximise immunoaffinity reactions. For the Column-IP, they followed pre-established and published protocols.

All samples were analysed in an LC-MS/MS system composed of an Easy-nLC 1200 HPLC system coupled to a Q Exactive HF-X mass spectrometer (ThermoFisher Scientific).

Data analyses

Data analyses

The authors adopted the data-independent acquisition (DIA) and the data-dependent acquisition (DDA) for analysing the resulting peptides. Library-directed DIA has been a widely used method for proteome and immunopeptidome analyses.

For the experiments using human B-cells, they built an RA957-specific spectral library in Spectronaut containing the DIA and DDA data from both the chip-IP and column-IP methods and used it to compare the results from both techniques.

However, due to the scarcity of human-derived samples, generating a spectral library based on DDA and DIA data from replicates is challenging. Thus, the authors created a library-independent “direct DIA”, a technique not yet applied to immunopeptidomics.

For non-canonical peptide identification, the authors combined information from Ouspenkaia et al. database on proteins translated from novel or unannotated open reading frames (nuORF) and the DIA data from their tissue samples.

Finally, the researchers created a generic spectral library, the “Lausanne Lib,” by combining the available raw MS files from previous samples measured in their laboratory with the UniProt and the nuORF database.

Materials:

  • Human B-cell line RA957
  • Snap frozen tumour tissues from patients with malignant melanoma and liver metastasis lesions
  • LSPone Laboratory Syringe Pumps
  • Microfluidic chip – manufactured at the Center of MicroNano Technology (CMi) at EPFL
  • 6 RVM Microfluidic Electric Rotary Valve
  • Easy-nLC 1200 HPLC (ThermoFisher Scientific)
  • Q Exactive HF-X mass spectrometer (ThermoFisher Scientific)
  • Spectronaut® software package (version 16.2, Biognosys)
  • NetMHCpan 4.1 prediction software
  • GraphPad Prism (version 9.1.0)

AMF - Xiaokang Li - Unleashing New Frontiers in Immunopeptidomics - AMF 4

Find the details of Materials & Methods here.

Results

AMF - Xiaokang Li - Unleashing New Frontiers in Immunopeptidomics - AMF 1

Microfluidic automation and Chip-IP enrich peptides from low cell numbers

The authors tested the sensitivity of the microfluidic automation chip-IP device to immunopeptidome enrichment using different low ranges of the human B-cell line, RA957, and HLA-I bound peptides. As a control, they ran the chip-IP in parallel with a column-IP system, using microbeads-packed columns and the same number of cells.

The resulting peptides from both methods were analysed in a Q Exactive H-X MS instrument using the data-independent acquisition (DIA) method. To compare both methods, the authors created a representative RA957 spectral library in Spectronaut containing the DDA and DIA generated from their experiments.

Although both methods generated a peptidome with the expected percentage of HLA-I binders and a 9mer-dominant profile, there was a drastic difference in peptide identification (ID). The microfluidic automation chip-IP method showed an unbiased enrichment in all conditions regardless of cell numbers, with a constant ratio of peptides mapped to the four main HLA-I alleles of RA957 cells and revealing more unique peptides.

Overall, the microfluidic automation with chip-IP technique showed:

  • Earlier elution profile
  • More peptides detected
  • Higher identification score with a lower number of cells, despite the lower MS2 intensities than in the column-IP

Overcoming insufficient sample supply

Xiaokang Li and colleagues used six tissue samples of various sizes (5-40 mg) from a malignant melanoma tumour for the chip-IP immunopeptidome enrichment. They analysed them using the same MS/MS setup used for the DIA method.

Insufficient sample supply, however, represents a significant challenge for the DDA and DIA methods that depend on acquiring data from multiple replicates for thorough data coverage. Thus, the researchers developed a library-independent approach—the “directDIA”—that is better suited for insufficient samples and has not yet been applied to immunopeptidomics studies (for more details on the directDIA method, refer to the original paper).

They found that more than 90% in four out of six tissues and 85% in the remaining two samples were HLA-I predicted binder peptides, showing the excellent purification performance of the microfluidic automation chip-IP system.

Notably, 12 tumour-associated antigenic peptides were found in the specimens from genes associated with melanoma: the MAGEA (melanoma-associated antigen) and the TYR (tyrosinase) gene, including some peptides known for their immunogenic capabilities.

The library search revealed a positive correlation between tissue weights and the number of identified peptides in each sample

Identifying non-canonical antigens

Standard protein databases usually list only canonically expressed molecules, which may hinder novel peptide discovery. The authors constructed a spectral library from peptides originating from novel or unannotated open reading frames (nuORF—REF) databases for the DIA data analysis of their tissue samples.

The approach revealed 554 peptides derived from non-canonical pathways, and 93% of those are predicted HLA-I binders – a critical observation in tumour profiling.

An immunopeptidome from both canonical and non-canonical sources was also observed when analysing samples from liver metastasis.

Enhancing antigen discovery

When comparing the different DIA analysis methods, they found that combining the Lausanne-Lib data set with either the UniProt or nuORF database enhanced peptide identification from tissues weighing more than 20 mg compared to the directDIA approach.

Conclusion - A versatile microfluidic automation platform for precision liquid handling

By integrating the IP and C18 peptide cleanup steps into one streamlined platform, the researchers reduced sample loss by cutting manual handling of all the intermediate and tedious sample transfer steps. The enhanced assay efficiency and sensitivity provided more peptide IDs using lower cell numbers.

We tested this new workflow for decoding the immunopeptidome from a low number of human cells and needle-biopsy-sized tumour tissues. We obtained excellent in-depth immunopeptidome compared to conventional format of microbeads-packed columns,” highlight the researchers.

While conventional column-IP assays use milligrams of antibodies per test, their new microfluidic automation chip-IP method significantly reduces it, making it a more economical assay approach thanks to the fully integrated fluidics workflow.

Developing automated and integrated sample preparation methods and better analytics tools in immunopeptidomics studies can significantly impact the future of therapy development in several clinical applications.

Non-canonical HLA-bound peptides are invaluable sources of potential tumour-specific neoantigens for effective immunotherapies,” remarked the authors.

Hundreds of non-canonical peptides were identified through the microfluidic automation chip-ID system, suggesting this new sample preparation technique using an integrated fluidic system is a powerful tool for future immunopeptidomics studies.
The programmable syringe pumps and the six-port switch electric rotary valve, controlled by Python scripts, can be used for different experiments and facilitate testing various conditions to optimise protocols more effectively. The OEM SPM Sequential Microdispenser is also a suitable alternative. It is for industrial applications with a tailored high-precision, low-volume rotary valve.

Sample preparation is a crucial step for immunopeptidome basic studies and clinical investigations. The authors’ designed microfluidic automation system reduced sample loss and the volume of reagents needed, improved sensitivity, and generated broader results with lower costs per assay.

References (AMA)

Li, X., Pak, H. S., Huber, F., Michaux, J., Taillandier-Coindard, M., Altimiras, E. R., & Bassani-Sternberg, M. (2023). A microfluidics-enabled automated workflow of sample preparation for MS-based immunopeptidomics. Cell Reports Methods, 3(6), 100479. https://doi.org/10.1016/j.crmeth.2023.100479

Made in Switzerland
10+ years of experience
Short Lead Time