Introduction
As we embark on missions to explore habitability and to search for signs of organic life in our solar system, the use of advanced and highly sensitive organic analysis instruments will be crucial (Neveu et al., Reference Neveu, Hays, Voytek, New and Schulte2018). Microfabricated Capillary Electrophoresis (CE) is a valuable asset in this quest, offering a proven track record in the analysis of trace organic materials on Earth and delivering rapid and high-resolution analyses for a diverse range of organics, varying in size, complexity, and charge and primary targeted functional group (Chiesl et al., Reference Chiesl, Chu, Stockton, Amashukeli, Grunthaner and Mathies2009; Hutt et al., Reference Hutt, Glavin, Bada and Mathies1999; Mora et al., Reference Mora, Stockton and Willis2013; Skelley et al., Reference Skelley, Scherer, Aubrey, Grover, Ivester, Ehrenfreund, Grunthaner, Bada and Mathies2005, Reference Skelley, Cleaves, Jayarajah, Bada and Mathies2006; Stockton et al., Reference Stockton, Tjin, Huang, Benhabib, Chiesl and Mathies2010, Reference Stockton, Tjin, Chiesl and Mathies2011). While this study is focused on amino acid samples, our functional group labeling approach (here targeting amines) is open to the detection of any primary-amine-bearing compound that might be present in a sample whether biotic or abiotic. The advantages and capabilities of CE are amplified by microfabricating the CE channels in glass substrates and integrating them with microfabricated fluidic networks to provide the added benefits of micro- to nano-liter scale volume fluid manipulation and sample analysis, automation, and lower limits of detection when coupled with laser-induced fluorescence (LIF) detection (Jang et al., Reference Jang, Razu, Jensen, Jiao and Kim2016; Kim et al., Reference Kim, Jensen, Stockton and Mathies2013, Reference Kim, Stockton, Jensen and Mathies2016). These attributes make microfabricated CE analyzers a valuable choice in the search for biosignatures in our solar system (Willis et al., Reference Willis, Creamer and Mora2015).
Significant efforts have been made to develop microfluidic CE prototypes that can be taken out of the lab for field and environmental testing. The first such portable microfluidic CE instrument, the Mars Organic Analyzer (MOA) prototype, was developed in Berkeley in 2005 (Skelley et al., Reference Skelley, Scherer, Aubrey, Grover, Ivester, Ehrenfreund, Grunthaner, Bada and Mathies2005). This prototype consisted of a microchip with pneumatic valves for basic solution control, a CE channel, a sensitive LIF confocal detection system, and electronics for operation and detection in a briefcase-sized portable package. The MOA prototype was successfully demonstrated in the field in the Panoche Valley, California where it detected amino acids from jarosite a mineral shown to be present in the Mars environment (Skelley et al., Reference Skelley, Scherer, Aubrey, Grover, Ivester, Ehrenfreund, Grunthaner, Bada and Mathies2005). A few years later this same instrument was successfully demonstrated in the Atacama Desert, Chile, detecting ppb levels of chiral amino acids from subcritical water extracts of the subsurface soils (Skelley et al., Reference Skelley, Aubrey, Willis, Amashukeli, Ehrenfreund, Bada, Grunthaner and Mathies2007). Subsequently, an analogous system coupled to a portable subcritical water extractor on a rover was deployed by others to analyze the levels of amino acids in the Atacama Desert (Mora et al., Reference Mora, Kehl, da Costa, Bramall and Willis2020). In these deployments, it was found that 10–100 picomolar analyte sensitivity (1–10 pptr) was needed for the meaningful detection of the dominant amine species in these samples namely, amino acids (Mathies et al., Reference Mathies, New, Golozar and Butterworth2021). The biological focusing of the amino acid composition and chirality on earth is a hallmark of its biological origin and analogous (but different) biochemical focusing might be found in other biosystems.
The feasibility of performing multiple CE separations in a varying gravitational field was first explored by Culbertson et al. (Reference Culbertson, Tugnawat, Meyer, Roman, Ramsey and Gonda2005) using a microfluidic platform with short preloaded 5-cm long glass CE devices. They successfully conducted repetitive gated CE injections and low-resolution 12-second separations of a mixture of four fluorescently labeled amino acids at high 2 µM concentrations during two zero-gravity parabolic flights. They observed nearly 24% variation in migration times between runs which was attributed to temperature variation. Additionally, their short channel lengths led to oscillating variations in migration times due to g-load variations as the aircraft transitioned in and out of the parabolas.
To advance beyond this pioneering work to a device and instrument that is appropriate for space flight and to address the limitations of the previous work, several questions must be addressed. Would longer channels mitigate the pressure sensitivity observed earlier with the added benefit of delivering the higher resolution and sensitivity needed to separate and identify a broader range of compounds at lower concentrations? This goal is also dependent on improved engineering that controls for thermal variation. Finally, there is a significant need for an automated flight design instrument, capable of conducting sample preparation and transfer on-chip, moving beyond preloaded samples.
Apparatus and methods
The Microfluidic Organic Analyzer (MOA) presented here is an advance on its earlier prototypes, that addresses the challenges posed above. MOA has been reengineered to Technology Readiness Level (TRL) 6 for the integrated system (system/subsystem prototype demonstration in a relevant environment), with a small number of components at TRL 5 (component validation in a relevant environment), at the Berkeley Space Sciences Laboratory and the University of Utah Department of Mechanical Engineering; the instrument has evolved significantly from the portable lab prototype introduced in 2005. The reengineered MOA utilizes a Programmable Microvalve Array (PMA) (Kim et al., Reference Kim, Jensen, Stockton and Mathies2013) for automatic fluid sample processing, a redesigned microfabricated 15-cm long CE separation channel for detailed analysis, and flight-format compact LIF detection to analyze picomolar concentrations and chirality of biomolecules such as amines, amino acids, and carboxylic acids (Casto-Boggess et al., Reference Casto-Boggess, Golozar, Butterworth and Mathies2022; Estlack et al., Reference Estlack, Compton, Razu and Kim2022; Estlack and Kim, Reference Estlack and Kim2022; Golozar et al., Reference Golozar, Chu, Casto, McCauley, Butterworth and Mathies2020, Reference Golozar, Casto, Estlack, Kim, Butterworth and Mathies2022). MOA’s capability to measure organic composition with a picomolar detection limit is likely necessary for any successful probe for habitability or biosignatures at the icy moons of Jupiter and Saturn (Mathies et al., Reference Mathies, Razu, Kim, Stockton, Turin and Butterworth2017, Reference Mathies, New, Golozar and Butterworth2021; New et al., Reference New, Kazemi, Spathis, Price, Mathies and Butterworth2021).
The flight-format MOA core integrated hardware underwent successful vibration and thermal testing followed by verification of its functional performance. Vibrational testing was performed on the integrated manifold, PMA-CE glass-PDMS wafer stack, and solenoids, as well as on the LIF detection system, following the guidelines laid out by the General Environmental Verification Standard for GSFC Flight Programs and Projects (2021; GSFC-STD-7000, https://standards.nasa.gov/standard/GSFC/GSFC-STD-7000). The system was tested along the X, Y, and Z directions at escalating levels of 1.7 Grms, 3.4 Grms, and 6.8 Grms. Detailed reports, specifically of the tests conducted at 6.8 Grms across three axes, have been provided in the SI. Furthermore, the PMA-CE wafer stack was found to be thermally robust. The glass CE wafers themselves are fabricated with a 668 C bonding process; the bonded PMA-CE glass-PDMS wafer stack underwent successful thermal testing at temperatures of ±50 C for a duration of 3 hours each.
A series of four Zero-G parabolic MOA flight tests was performed over a two-year period with a variety of goals: (1) ensuring that the support electronics, pneumatic valve drivers and operational programming were compatible with flight, (2) testing microfluidic pumping with the PMA and detection in variable g as reported already (Estlack et al., Reference Estlack, Golozar, Butterworth, Mathies and Kim2023), (3) performing a series of biomedical clinical analyses, and (4) performing CE separations under varied gravitational conditions that might be expected in space flight; one of the four flights was dedicated to this last goal. A Zero-G flight involves mounting and running the flight design MOA system in a specialized Boeing 727 aircraft. The aircraft then performs a series of parabolic flight arcs that simulate different levels of gravity, ranging from ∼0 to ∼1.7 g. A research format zero-gravity flight involves the execution of six sets of parabolic dives, with each set (∼5 minutes duration) consisting of five parabolas. The first set includes two Martian-gravity (∼0.38 g) and three Lunar-gravity (∼0.16 g) parabolas, while the remaining five consist of five microgravity (∼0 g) parabolas each. The flight also includes 25 s periods of hypergravity (∼1.7 g) per parabola as the aircraft is pulling out from the reduced-gravity portions of the flight. The parabolic sets are separated by brief periods of level flight, lasting approximately two minutes, to allow preparation for the next set of parabolas. A significant challenge is establishing the automation and concept of operations to effectively utilize the entirety of these flight profiles.
Figure 1a presents the portable apparatus developed for flight-testing of the MOA instrument. The instrument includes a 100-mm diameter PMA-µCE wafer, a manifold for fluid and pneumatic interface, a confocal LIF detection system, CE high voltage power supplies, and pressure and gravity sensor suite. Figure 1b shows top and bottom views of the core PMA-µCE-LIF organic analyzer unit. In May 2022, our first Zero-G flight was conducted and the functionality of the PMA and LIF detection system was successfully demonstrated (Estlack et al., Reference Estlack, Golozar, Butterworth, Mathies and Kim2023). The results showed that the microfluidic pumping, transferring, mixing, and degassing performed as well in different gravity conditions as on Earth under normal 1 g conditions. The positive outcome of this initial flight provided justification for a subsequent flight series with the goal of testing our ability to perform a variety of clinical assays, automatic on-chip sample transfer and CE separations under a range of extreme gravity conditions.
Portable apparatus for flight testing of the compact flight-format Berkeley Microfluidic Organic Analyzer (MOA) instrument. (a) Portable instrument includes PMA-µCE wafer, a fluid/pneumatic interface manifold, pneumatic control system, confocal LIF detection system, CE high voltage power supplies, Ground Support Electronics, sensor board, flow rate sensor, and a portable computer for data acquisition. (b) Top and bottom views of the flight-format core PMA-µCE-LIF organic analyzer unit. (c) Image of the flown microfluidic glass CE wafer produced at Berkeley Microfabrication Facility using standard photolithography and glass wet microchannel etching techniques (Golozar et al., Reference Golozar, Chu, Casto, McCauley, Butterworth and Mathies2020). The CE wafer features a folded cross-channel design with 110-μm channel width and 15-cm electrophoresis channel length, with 4.25-mm sample and waste arms etched isotropically 30 μm deep into a 100-mm diameter Borofloat glass wafer using hydrofluoric acid (49%). (d) Image of actual operation and monitoring of the MOA instrument in flight during a period of microgravity.
Results
In the flight on November 17th, 2022, we tested the ability of MOA to perform organic sample/biosignature CE separation and detection during varying gravity cycles (Figure 1d). The CE analysis was performed on the integrated glass CE chip, shown separately in Figure 1c. A mixture of proteinogenic and non-proteinogenic amino acids spanning a wide range of side-chain chemistries was fluorescently labeled and detected by LIF. A stock test mixture of model amino acids was made up by dissolving individual amino acids in purified water and then combining them to obtain a mixture consisting of 2 mM each of Arginine (Arg) and Histidine (His), basic; gamma-aminobutyric acid (GABA), non-proteinogenic; Valine (Val), branched chain; Serine (Ser), hydroxyl; Alanine (Ala), aliphatic; and Glycine (Gly), simplest; and 4 mM each of Glutamic Acid (Glu) and Aspartic Acid (Asp), acidic. Purified water was produced by first exposing 18.2 MΩ water (Direct-Q, Millipore Sigma) to a dual-wavelength UVC lamp (Black Magic 3D: UVC-25W-OZONE-WT) emitting at 254 and 185 nm for 12 hours. This was followed by distillation and an additional round of UVC irradiation to further reduce background amino acids. Subsequently, a 300 mM 10X borate buffer was formulated using this purified water by dissolving 1.43 g of sodium tetraborate (MW: 381.37 g/mol) into 50 mL water. These purification steps, together with sterile operation and procedural blanks run before and after each flight, resulted in blank control CE traces with minimal amine and amino acid contamination in all cases (SI Fig. 3); running an in-flight blank was not possible because of the shortened level-flight periods described below. A stock solution of XFD532 NHS ester fluorescent dye at 100 µM concentration (with excitation at 534 nm and emission at 553 nm) was used to label the amino acids at 200 nM each (400 nM for Glu and Asp) in a 30 mM borate buffer, with a pH of 8.5, and further diluted 5X into background electrolyte for CE analysis. The XFD532 labeling dye was used in place of our usual Pacific Blue because the fixed detection system was optimized for 532 nm excitation to facilitate the clinical assay evaluations. Prior to flight, the capillary channels (featuring a cross-channel design with a long 15-cm electrophoresis channel length, folded using hyperturn geometry to prevent racetrack broadening of the electrophoresis zones, and 4.25-mm sample and waste arms) were prepared by preconditioning with 1 M HCl for 10 minutes, then with purified water for 10 minutes, followed by 1 M NaOH for 20 minutes, and finally, they were flushed and filled with a background electrolyte that consisted of 30 mM borate buffer at pH 9.2. Also prior to flight, reservoir LV2 on the PMA was filled with 50 µL of the pre-labeled sample (see SI for PMA labeling).
Figure 2a-d schematically illustrates the sequence of fluidic and electrophoretic steps undertaken during flight to transfer a sample containing labeled amino acids into the sample well (∼80 s) followed by injection and electrophoretic separation (∼4–5 minutes). Figure 2a shows the transfer of 30 µL of sample from the LV2 reservoir to the CE sample well over the period of 80 s, using the PMA’s pneumatically controlled micropump. This transfer was done before the start of the set of parabolic maneuvers to ensure sufficient time for the separation and detection steps to occur during parabolas. After loading and as soon as the parabolic arc began, a 60-second pre-cleaning electrokinetic step was performed (Figure 2b), during which time a potential of −800V was applied to the sample and waste electrodes to retract any sample that had diffused into the cross-T injector back into the sample (and waste) well. The electrokinetic cleaning process took place in Martian-g and 1.7-g. Next, a 60-second cross-T electrokinetic injection was performed driving nanoliters of sample through the intersection by applying a −800 V potential to the waste electrode as shown in Figure 2c. During this 60 s operation, the gravitational potentials alternated between Martian-g and 1.7-g. Finally, a ∼2 min electrophoresis separation was performed with a 667 V/cm applied field (−10 kV at the cathode, −800 V back bias at the sample and waste, and ground at the anode; Figure 2d).
Illustration of sample handling and electrokinetic transport processes performed in flight to achieve the analysis of a mixture of amino acids. Blue denotes the background electrolyte buffer and red denotes the sample solution. Note in particular the indication of the different fluid heights in the various reservoirs as the analysis process proceeds. (a) Primed capillary channels with 30 mM borate buffer at pH 9.2 and sample transfer from LV2 on-chip reservoir to the CE sample well using the PMA’s pneumatically controlled micropump. (b) A 60-second pre-cleaning step (−800 V at the sample and waste electrodes) is used to retract any diffused sample in the arm back into the sample well. (c) A 60-second cross-T electrokinetic injection (−800 V at the waste) is performed to drive the sample through the intersection. (d) Electrophoresis separation with a 667 V/cm applied field (−10 kV at the cathode, −800 V back bias at the sample and waste, and ground at the anode). (e) Electropherogram of the amino acid biomarker CE separations under varying gravitational conditions, compared to the ground control analysis of the same sample (bottom trace) performed post-flight in a constant 1 g environment using MOA. The electropherogram presents the analysis of nine amino acids (40 nM each of Arg, His, GABA, Val, Ser, Ala, and Gly; 80 nM each of Glu and Asp). * indicate hydrolyzed dye peaks.
Figure 2e presents the results of a separation initiated during Lunar-g, transitioned at ∼40 s to 1.7-g for the majority of the separation, and transitioned at 80 s back to lunar-g. The separated molecules were detected 14 cm downstream of the injection point (5 mm before the cathode) using the flight-format confocal LIF detection system (Golozar et al., Reference Golozar, Casto, Estlack, Kim, Butterworth and Mathies2022). The top trace in Figure 2e presents the electropherogram of the successful separation of CE amino acid biomarkers, designated by their one-letter abbreviation, under these strongly varying gravitational conditions. The lower trace in Figure 2e presents the control analysis of the same sample performed post-flight in a constant 1-g environment (on ground). The in-flight electropherogram is superimposed on the gravitational force measured over the course of the experiment. Only 70 s of the flight, which is associated with the detection part of the CE analysis, are displayed in the results to clearly visualize the individual peaks of the amino acids. In both traces, the amino acid concentrations at the detection were 40 nM each of Arg, His, GABA, Val, Ser, Ala, and Gly; 80 nM each of Glu and Asp. The traces are normalized to 40 nM Arg injected concentration. In both the flight and control experiments, GABA and Val co-migrated, as did Ser and Ala. In the in-flight analysis, all amino acids except for the acidic species (Glu and Asp) were separated with an efficiency of more than 104 theoretical plates which equates to over 7.1 × 104 plates/m. While not as good as we previously reported using the dye Pacific Blue Succinimidyl Ester (Casto-Boggess et al., Reference Casto-Boggess, Golozar, Butterworth and Mathies2022), this result met the efficiency requirements for our instrument. The detection of the acidic species was unsuccessful because of the rising baseline after the second hydrolyzed dye peak in the flight analysis. This is due to leakage of the high concentration hydrolyzed dye into the injector during the run, which is likely caused by adsorption of the concentrated sample onto the sample channel walls that reduces the electro-osmotic flow (EOF) in the back biasing step. The comparison of the flight and ground control traces demonstrated agreement in peak intensities to within 5%, with a difference in migration times of less than 1%. This consistency aligns with the error observed in manual CE operations under a steady 1 g environment.
Based on our experience during the previous flights, for the CE evaluation we had planned to conduct six experiments and to use the level periods between the sets of parabolas to rinse the capillary channels for a minimum of 4–5 minutes to ensure consecutive successful runs. However, the level flight duration was unexpectedly shortened to ∼2 minutes each period. To accommodate this change, we reduced the number of experiments to three to allow for proper channel rinsing in between parabola sets. In the first experiment presented here, we observed a rising baseline after the second hydrolyzed dye peak. In these situations, a longer rinsing protocol of ∼10 minutes with running buffer ameliorates the problem, but due to time constraints during the flight, we proceeded with our original short rinsing protocol. This resulted in an increased baseline during the detection period in the subsequent run and a decrease in resolution between the His and GABA/Val peaks, as well as between the Ser/Ala and Gly peaks, during the second and third runs (with corresponding traces available in the SI). While the amino acids did undergo separation, the quality of the separation was not comparable with the first run, making the reporting of peak area variance unfeasible. The difference in migration times across the three consecutive flight runs ranged between 0.5% and 7%. The successful ground control trace was obtained post-flight and after the channels were rinsed for ∼10 minutes. This demonstrated that the rinsing variation was the origin of these differences, and this problem was uniquely caused by the unplanned altered flight protocol and is not intrinsic to our instrument or protocols.
Discussion
The operational difficulties encountered during the flight dedicated to CE evaluation are not unusual and demonstrate the necessity of having automated yet flexible operational protocols to optimize usage of the experimental periods. This adaptability is required in Zero-G flights and is also desirable for space flight. Despite these challenges, we were able to demonstrate the capabilities of the MOA platform in fluid transfer, cleaning, injecting, separating, and detecting samples with microfabricated CE under fluctuating gravity conditions. A significant concern was the effect of the enhanced gravitational potential during pull-out acting on the increased fluid column height at the cathode (see Figure 2d), which has the potential to alter the EOF during separation. The pressure difference generated by the height difference along the separation fluid column (∼400 µm) at the end of the separation was calculated to be 6.5 Pa under hypergravity conditions. This pressure difference is sufficient to cause a backflow equal to approximately 0.1% of the EOF (see SI for further details on these calculations) but, as confirmed by the experiment, this should not significantly disturb the separation quality. This result demonstrates that the MOA system is insensitive to the significant variations in gravitational potentials that could occur during space travel including changes in the pressure head caused by the acceleration and changes in pressure head for a MOA chip when a lander is tilted in a high gravity environment. Thus, the pressure head concerns raised in the early CE flight (Culbertson et al., Reference Culbertson, Tugnawat, Meyer, Roman, Ramsey and Gonda2005) have been addressed by our study along with the analysis of more complex mixtures and the demonstration of much higher sensitivity with a flight format instrument.
These conclusions rest on three in-flight separations performed during the single flight allocated to CE evaluation, one of which was completed with the nominal rinse protocol. The agreement of this separation with the ground control, together with the independently calculated 0.1% pressure-driven perturbation, supports our conclusion that large gravitational variation does not measurably degrade the separation. While it would also be desirable to perform chiral separations in-flight, which have been demonstrated with the MOA chemistry (Chiesl et al., Reference Chiesl, Chu, Stockton, Amashukeli, Grunthaner and Mathies2009; Skelley et al., Reference Skelley, Scherer, Aubrey, Grover, Ivester, Ehrenfreund, Grunthaner, Bada and Mathies2005), these separations require chiral-selector buffers and longer separation times that cannot be completed within a single parabolic gravity cycle.
Real planetary samples will likely present matrix effects such as high and variable ionic strength, extremes of pH, and divalent cations, all of which can degrade both separation quality and detection sensitivity. The MOA chemistry addresses these challenges with high-buffer-capacity background electrolytes, EDTA chelation of divalent cations, and sample dilution (Stockton et al., Reference Stockton, Chiesl, Lowenstein, Amashukeli, Grunthaner and Mathies2009). Each mitigation tends to reduce signal strength, but this is compensated in our approach by the picomolar sensitivity delivered by confocal LIF detection (Golozar et al., Reference Golozar, Casto, Estlack, Kim, Butterworth and Mathies2022), the improved fluorophore photophysics of Pacific Blue (Chiesl et al., Reference Chiesl, Chu, Stockton, Amashukeli, Grunthaner and Mathies2009), and optimized NHS-ester labeling (Casto-Boggess et al., Reference Casto-Boggess, Golozar, Butterworth and Mathies2022). Evaluating these effects under flight conditions remains an objective for future testing.
These successful results are an important step in advancing the TRL of programmable microfluidics and the picomolar sensitivity MOA CE system as a powerful probe for organic chemical signs of habitability and potential biosignatures. The advanced validation of our programmable microfluidics enables autonomous fluid processing and flexibility in operational concepts in flight operations as required here. The advanced validation of the MOA CE instrument brings high resolution but more importantly the picomolar sensitivity required for meaningful biosignature exploration (Mathies et al., Reference Mathies, New, Golozar and Butterworth2021) closer to deployment in actual space missions such as in situ screening for biogenic or abiogenic amines by rovers for sample return.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1473550426100391.
Acknowledgements
We thank the Berkeley SSL aerospace team Jeremy McCauley, Will Marchant, Miles Robinson, and Tim Quinn for the engineering of the MOA hardware, control, and detection optics. We also acknowledge Dr. Anna Butterworth and Claire Jenkins for assistance on the development and testing of the MOA instrument.
This flight activity was supported through the National Aeronautics and Space Administration (NASA) under grant No. 80NSSC21K0447 by SpaceTech–REDDI Tech Flights 2020 from the Space Technology Mission Directorate / Flight Opportunities. The development of the MOA instrument itself was supported by NASA through grant No. 80NSSC19K0616 issued by the SMD/Planetary Science Division through the ICEE-2 MatISSE Program.
Author contributions
Z.E., M.G. and J.K. designed the equipment, all authors designed the research and flight plan, M.G. and Z.E. performed the experiments, M.G. and R.M. analyzed the data, and all authors (M.G., Z.E., J.K. and R.M.) contributed to writing the paper.
Competing interests
The authors hereby confirm that they have no competing interests.
Open access
