Immunoanalytical Detection of Conserved Peptides: Refining the Universe of Biomarker Targets in Planetary Exploration

Abstract

Ancient peptides are remnants of early biochemistry that continue to play pivotal roles in current proteins. They are simple molecules yet complex enough to exhibit independent functions, being products of an evolved biochemistry at the interface of life and nonlife. Their adsorption to minerals may contribute to their stabilization and preservation over time. To investigate the feasibility of conserved peptide sequences and structures as target biomarkers for the search for life on Mars or other planetary bodies, we conducted a bioinformatics selection of well-conserved ancient peptides and produced polyclonal antibodies for their detection using fluorescence microarray immunoassays. Additionally, we explored how adsorbing peptides to Mars-representative minerals to form organomineral complexes could affect their immunological detection. The results demonstrated that the selected peptides exhibited autonomous folding, with some of them regaining their structure, even after denaturation. Furthermore, their cognate antibodies detected their conformational features regardless of amino acid sequences, thereby broadening the spectrum of target peptide sequences. While certain antibodies displayed unspecific binding to bare minerals, we validated that peptide–mineral complexes can be detected using sandwich immunoassays, as confirmed through desorption and competitive assays. Consequently, we conclude that the diversity of peptide sequences and structures suitable for use as target biomarkers in astrobiology can be constrained to a few well conserved sets, and they can be detected even if they are adsorbed in organomineral complexes.

The exploration of the solar system is gaining momentum as new technologies are facilitating missions to Mars and Jupiter’s and Saturn’s moons.¹⁻⁴ The main objective is to determine if life is or was present in these planetary bodies, as it would open new questions, such as where life originated or if it is an inevitable process given the right conditions. However, an unequivocal proof requires a mature technology sensitive and specific enough to avoid false positives or negatives and molecules that serve as reliable biomarkers. Regarding the technology, several systems are being developed, which need to be robust, sensitive, and fitted with space-suitable analytical techniques. In the 1976 Viking missions to Mars, the techniques consisted of pyrolysis or thermal volatilization of the sample. These were followed by gas chromatography coupled to mass spectroscopy, Raman spectroscopy, or fluorescence spectroscopy.^5,6 More recently, liquid-based techniques have been developed, like capillary electrophoresis, liquid chromatography, nanopores, and nanogaps that seek for a lower detection limit and for broadening the range of target biosignatures.⁶

The use of antibody-based biosensors has also been proven to be effective for astrobiological applications. The instrument SOLID3⁷ uses liquid extraction techniques to produce the soluble fraction of a sample,⁵ and its LDchip (composed of over 200 antibodies) interrogates for a range of bacteria and macromolecules (ca. > 1000 Da), including ancestral-like resurrected proteins,⁸ for immunofluorescence detection. As to the selection of biomarkers, it is still under debate whether the basic components of extraterrestrial life would have a different elemental composition of that of Earth, but it is a general assumption that it would be carbon-based.⁹ Macromolecules may yet vary from those we know, but studies agree that their building blocks (e.g., amino acids) can be chemically synthesized in early Earth conditions,^9,10 which implies that they are universal under similar biogeochemical conditions, as those present on early Earth or Mars.¹¹

Among macromolecules, peptides or lipids present inherent chemical stability even under extreme environmental conditions (e.g., radiation)^12,13 and for long periods of time within the geological record,^14,15 if compared to nucleic acids. While lipids give valuable information about their biological source,¹⁶ they require more sophisticated analytical techniques. Peptides, however, can be easily detected with antibodies, and though they are complex enough to be produced by prebiotic chemistry,¹⁷⁻¹⁹ they are simple enough to be ubiquitous in all living organisms. Determining the precise sequences composing the early peptides is almost utopic, as it is a blend of sequence and structure what has been selected over evolution.²⁰ The most conserved ones, relics of the early Earth, are called ancient peptides,²¹ and can be found in extant proteins because of the important functions that they still hold.²² Their ancestral function in the RNA–peptide world might have been related to the catalysis of simple metabolic processes, self-replication, or the binding of other molecules.²³ Later, they served as scaffolds for the folding and activity of more complex structures that led to actual proteins.²⁰ As a result, they are unequivocal and universal biomarker candidates and hold highly conserved amino acids that play a key structural or catalytic role, while other less conserved positions just fulfill certain chemical properties (e.g., hydrophobicity).

Detecting ancient peptides by sandwich immunoassays, like the LDchip, is hindered by their small size. Capture antibodies can detect them with high affinity, but the epitope gets embedded in the antibody structure, hindering the binding of the tracer antibody. Despite this, cortisol was unexpectedly identified in an environmental sample.²⁴ This could be explained by the establishment of noncovalent interactions with mineral particles of the soil,^25,26 resulting in decorated organomineral complexes than can be detected by the tracer antibody. In this work, we selected conserved ancient peptides of highly conserved enzymes, developed antibodies, and detected them using immunoanalytical techniques that can identify structural features over specific sequences. Adsorption of biomolecules to mineral particles could have occurred on early Mars as a preservation process of past life until our days.^27,28 We demonstrated that the detection of the peptides by sandwich immunoassays is enhanced when they are associated with relevant Martian minerals.

Experimental Section

Peptide Selection and Generation of Antibodies

Peptide sequences were selected from extremophile microorganisms living in habitats resembling early Earth conditions. The EMBOSS tool (ebi.ac.uk) was used for the alignments, with parameters BLOSUM 62, gap open penalty 5.0 (20 for HS-P), with an extend penalty 0.5; end gap open 10 and gap extend 0.5. Peptides were synthesized, and 100 μg of those were linked with a 1:1 ratio to the protein carrier Keyhole Lampet Hemocyanine (KLH) to immunize rabbits (Biomedal). The IgG fraction of blood serum was purified with protein A (HiTrap Protein A HP) and KLH (Sigma-Aldrich) as follows. KLH solution (0.1 mL at 50 mg mL^–1) was placed on a nitrocellulose membrane (0.22 μm GSWP, Millipore) in a Petri dish and dried at room temperature (RT, 1 h). 1 mL portion of concentrated (Amicon 30K) protein A-purified IgGs (2 mg mL^–1) was added on the membrane and incubated (2 h, 4 °C, dark) on an orbital shaker. The remaining fraction (∼800 μL) was collected and centrifuged (5000 g, 1 min), and the supernatant was collected into a new tube and quantified (Qubit). For fluorescent labeling, we incubated 100 μg of each purified antibody with 5 μL of 0.1 M NaHCO₃ pH 8.3 and 2 μL of Alexa Fluor 647 (Molecular Probes) in 50 μL of final volume (1 h, 300 rpm, dark). The mixture was added to a Sephadex 50-G column (Cytiva) and centrifuged (2000g, 2 min) to remove nonbound fluorophores. Labeling was measured at 280 and 650 nm, on a NanoDrop UV–vis spectrophotometer.

Printing of Antibodies and Peptide Microarrays

Antibodies and peptides were prepared at 0.8 and 0.4 μg μL^–1, respectively, in a protein printing buffer (PPB 1X, Sigma-Aldrich) with Tween20 0.02% for printing in microarrays. pH was shifted to 2 or 10 with HCl or NaOH, respectively, and measured with pH strips (Labbox-Export). Labeled and denatured protein A was used as a printing control and printed at 0.5 μg μL^–1, while PBS 0.1% Tween 20 (PBST) was used as a negative control in every array. Printing was done by using a three-spot pattern on epoxy activated glass slides (Arrayit) with a MicroGrid II TAS 600 arrayer (BioRobotics). The slides were left for 1 h at RT and stored (4 °C) until needed.

Immunodetection of Peptides

In direct assays, slides with printed peptides were blocked in Petri dishes first with 0.5 M Tris–HCl pH 9, 5% BSA (5 min, 300 rpm, RT) and then in 0.5 M Tris–HCL pH 8, 2% BSA (30 min, 100 rpm, RT). They were washed with PBST, dried by short centrifugation, and inserted into Hybridization Cassettes (AHC1 × 24, Arrayit). Slides were incubated with 50 μL of the antibodies at the indicated concentrations on an orbital shaker (1 h, 40 rpm, RT). After two washes with PBST, the hybridization chamber was dismounted, with the slide placed in a Petri dish for a third wash. Slides were dried and scanned (GenePix 4100A, Genomic Solutions) for fluorescence (635 nm) in the indicated photomultiplier tubes (PMTs) to optimize the dynamic range. Images were analyzed using GenepixTM Pro (Genomic Solutions), using the following equation to calculate fluorescent intensity (FI): FI = [(F635 Median-B)_sample – (F635 Median-B)_blank], F635-B being the median intensity at 635 nm minus the local background measured by the software. Statistical analyses were calculated using Prism 9.0 (Graphpad).

For sandwich immunoassays, blocked slides with printed antibodies were incubated with 50 μL of saturated organomineral suspensions (1 h, 300 rpm) in the hybridization cassettes. Wells were washed three times (5 min, 300 rpm) with PBST and incubated with 50 μL of the detection antibody (1 h, RT). Then, we proceeded as described above. For inhibitory immunoassays, concentrated peptides (400×), or PBS as control, were incubated with their respective labeled antibodies (30 min, RT, dark, 400 rpm). The resulting solution or the free peptide as positive control was incubated in direct assays to detect the printed peptides. Prism 9.0 (Graphpad) was used for statistical analysis.

Peptide Denaturation and Renaturation

Peptides or protein A as a printing control were denatured in SDS 2%, β-mercaptoethanol 10 mM (80 °C, 5 min). For renaturation, the buffer was filtered in G-50 Columns (Cytiva), with 45 μL of the denatured peptide added to the column for centrifugation (2000 g, 2 min). We added 25 μL of 1X PBS to the eluted peptide and checked the concentration (Qubit).

Mineral and Environmental Sample Processing and Analysis

A set of eight mineral samples from a mineral collection of the Centro de Astrobiologa (quartz, olivine, goethite, serpentine, albite, calcite, sphalerite, and hematite) and one obtained commercially (montmorillonite, Merck) were disinfected first by removing the superficial layer of the mineral with an automatic sander and then by immersing them in 5% KOH (15 min, 40 rpm). After rinsing in distilled water, minerals were immersed in 10% HNO₃ (15 min, 40 rpm), excepting calcite (only immersed in KOH). Minerals were then rinsed in water and ethanol (5 min), dried (50 °C, overnight), and then ground (FRITSCH Vibrating Cup Mill Pulverisette 9) to particles of 5–100 μm diameter using a tungsten mortar (21 s, 750 rpm). Mineral composition was checked by X-ray diffraction and scanning electron microscopy (INTA).

In the case of environmental samples, 0.5 g were suspended in 2 mL of buffer TBSTRR (Tris–HCl 0.4 M pH 8, NaCl 0.3 M, Tween20 0.1%), sonicated (3 pulses of 30 s), and centrifuged (2000 g, 2 min). As control, another 0.5 g of the samples were heated (500 °C, 16 h) to burn organics before sonication. Supernatants were used for immunodetection, as described above.

Peptide–Mineral Adsorption and Thermal Desorption

For adsorption, we incubated 20–75 μg of each peptide with 1.5 mg of mineral in 150 μL of PBS (2 h, 300 rpm). The suspension was then dried overnight at 42 °C and gently rehydrated in 150 μL of Milli-Q water. To measure the adsorption ratio, the complexes were formed using 83 μg of the peptides and 1 mg of the mineral. They were washed three times with Milli-Q water. After each wash, they were centrifuged (1 min, 14,000g), and peptide concentration in 20 μL of the supernatant was analyzed using Qubit (Thermo). For thermal desorption, the organomineral complexes were heated (300 °C, 16 h) using bare minerals as controls.

Results and Discussion

Identification of Ancient Peptides as Potential Biomarkers

When seeking life out of Earth, it must be considered that the necessary macromolecules may differ from those that we know. However, the basic components of life must be somewhat conserved, as they are products of complex biogeochemical reactions that occur under specific conditions. As biomarker candidates, we sought to identify highly conserved peptides from the literature that can acquire their secondary structure autonomously and exhibit functional features.²¹

We selected seven protein motifs (Table 1) proposed to harbor ancient peptide folds^29,30 of the RNA–peptide world.^31,32 They serve as a scaffold for the efficient folding of proteins, like β-propeller (β-prop), and/or stable interactions with molecules to enable metabolic processes [helix-hairpin-helix (HhH), β-Hammerhead (βHH), P-loop, 4Fe4S, Rossmann (Ross), and TIM-barrel (TIM)]. Some show a highly conserved sequence pattern within all protein families, such as P-loop, while others (βHH, TIM, β-prop) share a common folding but show more variation between families. From each sequence, we delimited peptides of 15–47 amino acids length that still retain the properties of the motif and picked representative sequences from bacteria (Figure 1a) showing a similar metabolism to that considered to be present in the origin of life.

Table 1. Ancient Peptides Used in This Study.

motif	binding interactions	model peptidea
HhH	phosphates (P) of DNA and RNA	LLELVRIRHIGRVRARKLYNAG
P-loop	P of nucleotides	LVLGIDGLSRSGKTTLANQLSQT
Ross	P of (di)nucleotides and some coenzymes	DSVLIVGAGPSGSEAARVLMESGYTVHLTD
TIM	metals, P, catalytic domain, metabolites	DKDEAWKQVEQLRREGATQIAYRSDDWRDLKEAWKKGADAILLIDAT
βHH	RNA	RLLVKDGDYVEAGQPLTR
β-prop	ions, metabolites, protein–protein interactions	KPHALLWGPDNQIWLTERATGKILRVNPESGSVKTVF
4Fe4S	cluster Fe4S4	ENCIQCNQCAFVCPH

^aConserved positions of the motifs in bold and underlined.

Figure 1.

Folding prediction of the model peptides. (a) Peptide structure in the parental protein, indicated by their accession number (PDB) and location within the protein, showing ligands for HhH (single-stranded DNA), Ross (adenosine diphosphate), β-prop (glycerol), and 4Fe4S (4Fe4S cluster). (b) Most probable predicted structures of the model peptides in aqueous solution.

We then modeled the folding of the peptides with PEP-FOLD 3.5,³³⁻³⁵ which calculates the five most probable structures expected from a specific sequence (Figure 1b and S1). The most frequent structures coincided with folding within the parental proteins in at least one representative prediction. For HhH, all predictions coincided, also suggesting that it folds autonomously, and for the β-prop two most probable predictions completely fitted the parental x4-stranded β-meander. The spatial distribution of all Ross predictions coincided with the structure within the protein, although the expected (βαβ) showed no secondary structure in 60% of the models. Similarly, for the P-loop, a helix structure was predicted, but the β-strand of the parental subunit (βα) did not appear. TIM simulation also missed two β-strands from the original fold (βα)2, but 80% of P-loop predictions coincided with the spatial direction of the peptide backbone. Folding predictions of βHH and 4Fe4S exhibited a more disorganized structure, despite showing a similar spatial distribution, but altogether, we considered that selected peptide sequences would autonomously acquire their expected folding even when they were isolated from their flanking sequences.

Polyclonal Antibodies Target the Secondary Structure of Ancient Peptides

Considering that different peptide sequences in early Earth or even in other planetary bodies could share a common secondary structure, we aimed at determining whether polyclonal antibodies would detect similar folds, even with low sequence similarity. We tested the specificity and cross reactivity of the generated antibodies by direct fluorescent immunoassay against the immobilized peptides (Figure 2). We estimated the affinity of each antibody by incubating serial dilutions, and the optimal concentration ranged from 20 to 130 nM. Α-βHH, α–β-prop, α-P-loop, and α-TIM showed the highest affinity, while α-HhH, α-Ross, and α-4Fe4S required higher concentrations of the antibody. Additionally, α–β-prop cross-reacted with βHH at 61% of the total signal, and α-P-loop and α-4Fe4S also recognized HhH at their working concentration, pointing out that the antibodies may be detecting a common epitope on the peptides that could depend on the secondary structure.

Figure 2.

Characterization of polyclonal antibodies against ancient peptides. (a) Antibody affinity was tested by direct fluorescent immunoassay by doing serial dilutions of the purified antibodies. Dot lines indicate the selected work concentration. Blue line: specific binding was with the Hill slope regression model. The X axis is log(concentration). Error bars represent the standard error of the mean (SEM). Exposition at PMT = 800, except for α-4Fe4S (PM = 900) (n = 2 arrays) (b) Nonspecific signals (cross reactions) identified at the working concentration of the antibodies, represented by a percentage of the maximum signal obtained for each antibody. Exposition at PMT = 800, except for α-4Fe4S (PMT = 900) RFU: relative fluorescence units.

To test whether the epitopes were linear or three-dimensional, we performed direct fluorescent immunoassays using the peptides under normal or denaturing conditions (Figure 3). Under denaturing conditions, all antibodies except α-TIM detected their respective antigens with a significantly lower signal, while α-TIM also showed the same tendency. This suggested that the detection of these peptides is highly dependent on the specific folding. To test this hypothesis, we renatured the peptides by incubating them in their original buffer, followed by detection using a direct immunoassay (Figure 3). We observed that βHH and β-prop were detected with the same intensity as the non-treated peptides, and the α-TIM signal was also increased, suggesting that they recognize a conformational epitope that can fold autonomously even after denaturation. Otherwise, HhH, 4Fe4S, P-loop, and Ross were not detected in renaturing conditions, maybe because it affects these peptides differently. In an independent experiment, we also incubated the peptides with a mix of all the antibodies (Figure S2), observing similar results except for the detection of HhH in denaturing conditions, a lower signal in P-loop, and the detection of Ross in renaturing conditions. This may be explained by the Hook effect, nonspecific detection by other antibodies, or the recognition of newly accessible antigens after the treatment.

Figure 3.

Influence of the peptide structural features for antibody detection. Non-treated (−), denatured (D), or renatured (R) peptides were tested in direct immunoassays. Error bars represent SEM (n = 3, PMT = 700). C: denatured and labeled protein A (Alexa 647). RFU: relative fluorescence units. Data were analyzed using one-way ANOVA with Dunnett’s post-test comparing means with the mean of denatured condition. **P < 0.01; ***P < 0.001.

Polyclonal Antibodies Target Dissimilar Peptides Sharing a Common Folding

Given that the peptide structure, rather than the linear disposition, was being recognized by the antibodies, we explored if the antibodies would recognize peptides with different amino acid compositions but predicted to fold as the original ones. We selected peptides with a varying homology from the original sequence but preserving the conserved pattern sequence (Table 2 and Figure S3). Additional homologous sequences were synthesized for HhH and 4Fe4S due to their high dependence on the folding seen above. For these peptides, we also designed a hypothetical sequence (HS-H6, HS-F6) resembling the same folding by substituting each amino acid for another, showing similar properties (e.g., Glu [E] for Asp [D]).

Table 2. Selection of Homologous Sequences for the Model Peptides.

motif	ID	alternative sequencea	identityb	organism	refc
4Fe4S	HS-F1	EHCIQCNQCAYVCPH	86.7%	Clostridium culturomicium	WP_040194890.1 (686–700)
	HS-F2	DVCVQCGKCVMVCPH	53.3%	Chroococcidiopsis thermalis	K9U6D6 (726–740)*
	HS-F3	SVCVGCRYCMVACPF	33.3%	Desulfococcus multivorans	S7TUH9 (102–116)*
	HS-F4	LNCRHCEKAPCVEVCPT	41.2%	Thermococcus gammatolerans	C5A2R4 (46–72)*
	HS-F5	ENCIQCKYTDCAEVCPV	64.7%	Acidithiobacillus thiooxidans	A0A1C2I2N3 (7–23)*
	HS-F6	DQCLNCQNCVIACPK	33.3%		This work
HhH	HS-H1	STDIQYAKGVGPNRKKKLKKLG	24%	Thermotoga maritima	1GM5 (A:114–135)
	HS-H2	QLRIEDFYGVGPATATKMKSLG	28%	Synechococcus sp.	B4WH69 (194–215)*
	HS-H3	LRFLEIPGLGPKKAKIIYET	22.7%	Thermosulfurimonas dismutans	A0A179D669 (91–110)*
	HS-H4	ELDIADVPGIGNITAEKLKKLG	31.8%	Saccharolobus solfataricus	1JX4 (A:177–198)
	HS-H5	IEELSKLPGIGPKTAQRLAFFI	29.2%	Caldanaerobacter subterraneus	3VDP (A:25–46)
	HS-H6	IVDVIKLKGVGKAKAKRLVLAG	22.7%		This work
βHH	HS-BH	KILVKEGDTVKAGQTVLV	52.6%	Propionibacterium freudenreichii	1DCZ (A:67–84)
β-prop	HS-Bp	GAVISPDGTKVYVANAHSNDVSIIDTATNNVIATVPAG	16.7%	Methanosarcina mazei	1L0Q (A:36–73)
TIM	HS-T	DVDEMLKQVEILRRLGAKRIAVRSDDWRILQEALKKGGDILIVDAT	68.1%	de novo sequence	7P12 (A:2–47)
P-loop	HS-P	VGVLITGDSGIGKSETALELIKR	26.9%	Staphylococcus xylosus	1KO7 (A:145–167)
Ross	HS-R	KNVVIIGLGLTGLSCVDFFLARGVTPRVMD	26.7%	Escherichia coli	2JFG (A:7–36)

^aMotif conserved sequence typed in bold and underlined.

^bSequences identity of the pairwise sequence alignment versus the model sequence, using the Needleman–Wunsch algorithm (see methods).

^cProtein Data Bank (PDB) and Uniprot (*) references. Ref. HS-F1 from the NCBI.

Although anticipating lower detection signals, we tested whether the antibodies would recognize the homologous peptides by direct immunoassays at neutral pH. Having the original peptide as a positive control, we observed that though the affinity toward the new peptides was reduced, α-HhH and α-4Fe4S showed a positive signal when incubated with their respective alternative sequences, except for HS-F4 and HS-F5 (Figure 4). For HS-H1 and HS-H6, we observed a level of detection similar to that of the model peptide and similar to that of HS-F2 and HS-F3. In the case of HS-F1, the signal was even higher than that of the model peptide, reinforcing that the antibodies can recognize structural features of the peptide over the specific primary sequence. However, binding of the antibodies to the homologous sequences to βHH, Ross, β-prop, P-loop, and TIM could not be detected.

Figure 4.

Detection of the homologous peptide sequences is described in Table 2. Direct fluorescent immunoassays were performed with peptides incubated at pH 2 (orange), pH7 (dark blue), or pH 10 (cyan). Error bars represent SEM, (n = 3). PMT = 700, except for 4Fe4S and HhH (PMT = 900). RFU: relative fluorescence units.

To determine whether a change in the amino acid charges had an effect on the recognition by the respective antibodies, we shifted the pH of the peptide solutions to 2 and 10. At pH 10, the detected signal was higher for the 4Fe4S model peptide, as well as for its homologous peptides HS-F1, HS-F2, and HS-F6, pointing to an influence of the pH for the immunodetection. At pH 2, HS-F2 and HS-F3 were detected with a higher signal than at pH 7, being the optimal conditions for the detection of HS-F3. Incubation of peptide HhH at pH 2 or 10 showed positive signals, but these conditions did not improve the detection. Regarding the other peptides, only a faint increase in the detection was observed for HS-BH and HS-R at pH 10. Given the observed cross reactions (Figure 2b), in an independent experiment, we incubated the peptides with a mix of all the antibodies, resulting in an increase in the detected signals for peptides HhH, 4Fe4S (Figure S4). HS-BH and HS-R could be now detected at pH 10 with a higher signal, but still the antibodies did not bind HS-Bp, HS-P, or HS-T. In summary, those antibodies showing a lower affinity against the original peptide recognized better alternative sequences exhibiting a common folding.

Ancient Peptides Adsorbed to Minerals are Detected by Sandwich Immunoassays

Small molecules, including peptides, cannot be detected by sandwich immunoassays because they get embedded in the binding site of the capture antibody, and no free epitopes are available for binding to the tracer antibody. The identification of small molecules, like cortisol,²⁴ in the soil led us to hypothesize that the peptides bind to the surface of mineral particles or other particulate material to form organomineral complexes. From an astrobiological point of view, we tested this hypothesis with minerals and rocks described to be abundant on the Martian surface.³⁶ Their composition was analyzed by X-ray diffraction, (Table S1) and to explore nonspecific recognition by the antibodies, we performed sandwich immunoassays using the bare minerals at saturating conditions (Table S2). Unexpectedly, montmorillonite was indistinctly detected by all antibodies, while calcite, serpentine, and albite were identified with varying signals depending on the antibody, observing lower interactions with α-TIM, α–βHH, and α–β-prop. In the case of quartz, olivine, goethite, sphalerite, and hematite, nonspecific interactions with the antibodies were observed to a minor extent.

Hence, we selected peptides βHH, P-loop, and TIM to produce complexes with those minerals showing less unspecific antibody binding. We first tested the efficiency of the adsorption of these peptides to each mineral by measuring the nonbound peptide concentration after washing the organomineral complexes (Figure S5). It showed minor differences among the minerals (40–60% adsorption), except for montmorillonite (100%). Then, we ran sandwich immunoassays, incubating the bare minerals or the free peptides as controls (Figure 5a). Importantly, while the free peptides were not detected using this technique, after adsorption of some minerals, they could be identified in the sample. We observed a significantly increased signal when α–βHH was incubated with βHH complexes based on quartz, olivine, and sphalerite but not with goethite, serpentine, albite, or hematite. P-loop detection was significantly increased when adsorbed to olivine, sphalerite, and goethite with a similar trend. Regarding TIM, the signal was significantly increased when adsorbed to quartz, olivine, and serpentine but also showed an increased signal when bound to goethite. These results showed that under the right conditions, peptides can form organomineral complexes that enable their detection by sandwich immunoassay.

Figure 5.

Detection by sandwich immunoassays of ancient peptides as a part of organomineral complexes. (a) The free peptides (F), buffer (C), bare minerals, as indicated, or organomineral complexes constituted by selected minerals and peptides βHH (n = 4 for quartz and sphalerite; n = 6 for olivine; n = 2 for goethite and hematite; PMT = 800), P-loop (n = 3, PMT = 800), or TIM (n = 4 for quartz, olivine, goethite and serpentine; n = 3 for albite, spharelite, and hematite; PMT = 500) were tested. Error bars represent SEM. One-way ANOVA with Dunnett’s post-test with the mean of bare mineral as the control. (b) Specificity of the signal observed by treating the bare minerals of the organomineral complexes with heat (PMT = 500). (c) Sandwich immunoassay of the free peptide (F), buffer (C), bare olivine (Oliv), or organomineral complex (TIM-Oliv). A competition assay was performed by preincubating the specific peptide (TIM) or a control peptide (βHH) with TIM-Oliv (PMT = 600, n = 1). (d) Sandwich immunoassay of environmental samples BB, VHT, and VHB, developing with the antibody mix (PMT = 900). RFU: relative fluorescence units. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

To confirm that the detected signals were specific to the peptide bound to the mineral and not an artifact, we inactivated the peptide structure by heating the sample (300 °C). Using TIM as a model antigen, the organomineral complex was heated along with the bare mineral (Figure 5b). As expected, TIM complexes with quartz, olivine, and serpentine were detected, while the signal was lost after thermal desorption. However, in the case of goethite, heat treatment altered the mineral in a way that the bare mineral showed a higher signal than the heated complex. Thus, by heating the complex, the degradation of the peptide was masked. Lastly, to confirm that the observed signals were specific for TIM, we performed an inhibitory immunoassay by pre-incubating free TIM or βHH with α-TIM antibody prior to the sandwich immunoassay with the TIM-olivine complex as the antigenic sample (Figure 5c). Preincubation with TIM resulted in a lower detection signal, suggesting that neither the free peptide nor the detached peptide were detected. As expected, when βHH was used in the preincubation, the signal was not affected, confirming that the detection of TIM is due to the specific binding of the α-TIM antibody to TIM on the mineral complex.

Finally, to check whether some of the α-peptide antibodies developed herein could recognize any specific signal in natural environmental extracts, we tested by sandwich immunoassay the three independent samples (BB, VHT, and VHB) from the hydrothermal field of El Tatio (Atacama, Chile), as an analogous environment of the Mars surface, in which ancestral proteins have been shown to be present.⁸ We included a sample of an Antarctic biofilm that lacks soil material as a control (Figure 5d). All samples were also heat-denatured and used as blank. We used the antibody mixture as tracer antibodies and subtracted the fluorescence signal of the heated samples. Results showed low but positive signals for β-prop, P-loop, 4Fe4S, Ross, and TIM in all samples from El Tatio, while βHH was not detected, and HhH was detected just in BB. No signal was observed from the biofilm.

All in all, we obtained polyclonal antibodies that recognize a set of seven relevant peptides for life detection in planetary exploration, observing that some can identify specific folds even with varying amino acid sequences. If the peptides are adsorbed to minerals, they behave as a multiepitope complex and can also be detected by sandwich immunoassay, confirming the feasibility for the identification of other relevant small molecules preserved in the geological record.

Conclusions

Planetary exploration requires the identification of universal and unequivocal biomarkers and the development of reliable techniques to determine the presence of living organisms or life remnants. In this study, we evaluated the detection with antibodies of ancient peptide folds, which exist in all forms of life and are believed to have been present in the RNA-peptide world. We have also determined that their detection can be enhanced if adsorbed to soil minerals, an association that also preserves the peptides from decay.^26,37,38

The search of universal biomarkers is controversial, as some biomolecules like short peptides or short chain fatty acids can be synthesized abiotically and are found in meteorites.³⁹⁻⁴¹ Yet, they have random sequences and a limited range in their amino acidic composition. Instead, there is an increasing consensus regarding ancient peptides as products of biochemistry as their folding patterns are universal and have been selected during evolution for a specific function, situating them at the interface of prebiotic and biotic systems. We considered ancient peptides β-prop, HhH, βHH, P-loop, 4Fe4S, Ross, and TIM because of their function in extant proteins—they interact with specific and essential molecules for life—, their wide distribution in different protein families and their composition, which is enriched in amino acids and thought to be present in the prebiotic world.^42,43 The αβα sandwich structure has been consistently identified as the oldest architecture,⁴³ but folding alone does not confer an interacting role. Similarly, helices are a common structural element in proteins, but only specific tandem helices with characteristic sequence patterns, like HhH, establish interactions with the DNA.⁴⁴ It is the combination of folding and the conservation of a minimal pattern in their sequence that distinguishes these peptides and enables interactions that confer their functionality. Precisely, peptides that bind metal clusters, such as 4Fe4S, have conserved Cys residues in their sequence and mediated electron transfer in biochemical reactions, facilitating and regulating the energy flow of metabolic pathways. The P-loop is characterized by binding phosphate groups and is present in almost every protein using ATP as the substrate, and it is believed to share a common origin with Ross and TIM, and even other relevant folds, like the Flavodoxin-like, HUP-like, and HAD-like.⁴³

Bioinformatic tools predicted that the peptides fold autonomously, acquiring a structure similar to that found within their respective parental protein, as expected from their ancient origin. The polyclonal antibodies let us confirm the prediction when we exposed the peptides to denaturing conditions, like those present on early Earth (high temperature, reducing conditions), and detected a lower or even no signal (Figure 3). Autonomous folding depends on various solvent factors, and it is reasonable to assume that not all peptides possess the ability to refold into a stable structure, but βHH, β-Prop, and TIM could refold after denaturation, supporting their importance in early life, where autonomous folding of peptides preceded the formation of proteins. Nonetheless, the main aim of this work is to demonstrate whether polyclonal antibodies generated against a specific fold can also identify dissimilar sequences. Antibodies α-HhH and α-4Fe4S showed lower affinity than others, but they were fully dependent on the structure of the peptide and proved to be more promiscuous in the detection of sequences that share the same structure. The detection was enhanced in some of the sequences homologous to these two peptides when the pH was changed, suggesting that ionic interactions play a role in the specificity of the antibodies. This was confirmed when a hypothetical sequence of HhH or 4Fe4S sharing the electrostatic interactions with the original peptide was also detected. It is important to note that extreme environments with low pH like Rio Tinto (Spain) are considered Mars analogues.⁴⁵ For those folds whose homologous sequences were not detected, a different folding pattern may be explained, and more sequences could be evaluated. Additionally, some antibodies exhibited cross-reactivity with other nonpaired peptides, but this remains plausible because of the recognition of common super secondary features shared by the different peptides, like an α-helix followed by a loop (P-loop and HhH) or the presence of β-sheets (βHH and β-prop). Considering that the main objective is the detection of structures as life biomarkers, antibodies showing cross-reactions will broaden the detection capabilities of the technique.

Following previous observations of small molecules being detected by sandwich immunoassays in environmental samples,^24,46 we wanted to shed some light on this phenomenon, considering that the ancient peptides could just be detected with this technique at low levels (Figure 5), probably because they get embedded in the capture antibody. Quartz and hydrated silica have been detected in Martian sediments, particularly in the Valles Marineris region and in ripples and dunes within Antoniadi and Isidis basins.^47,48 Olivine and Mg-rich olivine rocks have been identified in various volcanic regions, including Syrtis Major, Gusev Crater, and on the eastern floor of Kunowsky crater.⁴⁹ Goethite has been observed in association with sulfate-bearing deposits, such as those found in Valles Marineris, Meridiani Planum and in the Columbia Hills at Gusev crater.⁵⁰⁻⁵² Serpentine minerals have been detected in several locations including the Claritas Rise region, along the Thaumasia highlands to the south of Valles Marineris and in the Nili Fossae region.⁵³ The formation of organomineral complexes has been described in the soil during wet–dry events,⁵⁴ where the aqueous phase containing the dissolved molecules evaporates if exposed to high temperatures. These peptides may withstand such temperatures during the adsorption process, maintaining their fold. For instance, the selected TIM sequence corresponds to a fragment of a de novo sequence⁵⁵ that was calculated to have a melting temperature of ∼88 °C. Hence, they are stable structures that may have been adapted to the high temperatures of the early oceans.

Regarding the adsorption of the peptides to the minerals, differences in the efficiency of establishing noncovalent interactions may both depend on the chemical and physical properties of the peptide and the mineral themselves, such as surface, porosity, and chemical composition. For example, basic polyamino acids usually show greater adsorption than acidic polyamino acids on goethite and montmorillonite, and vice versa.⁵⁶ Nevertheless, demonstrating that organomineral complexes can be detected with sandwich immunoassays implies that instruments based on this technology, like SOLID3-LDChip, possess a wider range of detection than expected. Besides, some bare minerals could confound some antibodies, as that happened with montmorillonite, a phenomenon that should be further explored in the future. Other unspecific signals may be due to the saturating concentration of the peptides used for the adsorption, rather than unspecific binding of nonrelated molecules or peptides, as demonstrated in the inhibition assay or the thermal treatment of the minerals. As a result, adsorption protocols will be refined by reducing the inorganic fraction and adding reagents that limit nonspecific binding of the antibodies to the mineral fraction.⁴⁶ Nonetheless, there is a lower concentration of peptides in the environmental samples, as observed. Although weak, the positive detection of some of these peptides in the samples of a Mars analogue environment demonstrates the power of this technology.

In summary, the use of ancient peptides is a promising strategy in the search for life on other worlds, particularly Mars. These peptides, being key in both the origin and the development of life, were detected in environmental samples through their interactions with minerals. The set of antibodies generated in this work could be implemented in SOLID-LDChip for the identification of ancient peptides in environmental samples, as these and other biomarkers forming organomineral complexes could greatly help in the quest to find life elsewhere.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.3c04165.

• PEP-FOLD3.5 structure prediction of the ancestral motif sequences in aqueous solution at neutral pH; influence of the model peptide structural features in the detection using a mix of the antibodies; PEP-FOLD3.5 structure prediction of selected homologous sequences; influence of the pH in the detection of homologous peptide sequences described in Table 2 using a mixture of the antibodies; mineral composition determined by X-ray diffraction; and nonspecific interactions of bare minerals at saturating conditions with the antibodies in sandwich immunoassays (PDF)

Author Present Address

^§ Centro de Biología Molecular Severo Ochoa UAM-CSIC, Madrid 28049, Spain

Author Contributions

D.R.-G. and V.P. designed the study. D.R.-G. designed the experimental scheme. P.M.-d.-S. conducted the experiments. D.R.-G. and P.M.-d.-S. wrote the manuscript, which was revised by V.P. V.P. supervised the project. All authors edited and have given approval to the final version of the manuscript.

The authors declare no competing financial interest.