There is some indication that Mars was warmer and wetter in earlier periods of its history than it is today.^[3–5] Since the current upper surface of Mars is exposed to irradiation and oxidation,^[6] any life form, if present, might have retreated into protected regions of rocks and minerals,^[7] because on the surface any biological substance would likely become destroyed. Therefore, the main purpose of sampling with the aid of a drill is to penetrate through the surface into deeper layers, below the assumed zone of damage to biomolecules.

Different detection methods were proposed for these type of missions, which would be applicable, besides the investigation on Mars, also to other extraterrestrial targets, including future return samples^[8,9] – whose acquisition is being reconsidered^[10,11] – e.g.: (1) extraction of putative biomolecules and subsequent immuno assays, whereby the approach capitalizes on the flexibility, sensitivity and specificity of antibody-antigen interactions;^[12] (2) probing for morphological biosignatures due to bona fide microbial fossils as well as microbially influenced sedimentary structures;^[13] or (3) Raman spectroscopy for the detection of both extinct and extant microbiota, since this technique is sensitive to organic and inorganic compounds^[14] and able to unambiguously identify key spectral markers in a mixture of biological and geological components.^[15–19] The latter method also has the advantage of being a non-intrusive in situ analysis method, requiring very little effort in preparing the samples (e.g. no chemical extractions are necessary), and compact instrumentation has already been developed.^[18] The potential for the identification of microorganisms and biomolecules has been recently reviewed.^[20–22]

On Earth, high concentrations of salt are not a deterrent for many types of microorganisms – on the contrary, extremely halophilic archaea (now called haloarchaea), bacteria and some unicellular algae thrive in saturated brines.^[23] Their optimum growth occurs in media containing 2.5–5.2 m NaCl;^[24] their natural environments are the Dead Sea, the Great Salt Lake, sabkhas and natural or artificial salterns.^[23] Ancient salt deposits are another habitat and viable haloarchaea were isolated from salt sediments of Permian and Triassic age.^[25–28] Haloarchaea were demonstrated to accumulate in the tiny fluid inclusions of halite^[29,30] and they may have survived there for long periods. Therefore, the possible survival of haloarchaea over geological times,^[31] combined with the detection of Martian halite, make the search for halophilic microorganisms on Mars plausible.

Martian environments regarded as promising locations for the search for life include paleolake craters and evaporitic deposits.^[5,15,16] Elements from Martian soil and rocks that were determined with the alpha-particle X-ray spectrometer included Na, Mg, Cl and Br.^[32] Evidence for halite was found in the shergottite, nakhlite, and chassigny (SNC) meteorites,^[33] which are of Martian origin.^[32,34] Saturated salt solutions, which would possess greatly depressed freezing points,^[35,36] could be also envisaged on Mars; they may not be present as large pools, but rather could occur in tiny pore spaces between mineral grains.^[36] Very recently, large chloride deposits were found on Mars by analysing data from the Mars Odyssey Thermal Emission Imaging System (THEMIS).^[37]

UV radiation exposure is one of the most important stresses encountered by halophilic microorganisms^[5] in their natural environments and therefore they are thought to have developed carotenoids as protectant molecules. In a recent experiment Halococcus dombrowskii H4 DSM 14522^T was exposed to ultraviolet (UV) doses over a wavelength range between 200 and 400 nm, simulating Martian UV flux.^[38] Halite-embedded cells showed no loss of viability following exposure up to about 21 kJ/m² and the estimated dose of 37% survival (D₃₇) was ≥400 kJ/m². However, exposure of cells to UV flux while in liquid culture reduced D₃₇ by two orders of magnitude (up to about 1 kJ/m²). Similar results were obtained with Halobacterium salinarum NRC-1 and Haloarcula japonica. The absorption of light by colour centres resulting from defects in the crystal structure of halite was likely contributing to these results. Under natural conditions, haloarchaeal cells become embedded in salt upon evaporation; therefore, dispersal of potential microscopic life within small crystals, perhaps in dust, on the surface of Mars would resist damage by UV radiation.

In contrast to bacteria, archaea lack several characteristic biomolecules such as fatty acids, murein and lipopolysaccharide (Ref. [39] for a recent review and Ref. [40] as online source of information), but possess some other unique compounds, e.g. large amounts of carotenoids in the case of haloarchaea, the principal ones being C₅₀ straight-chain α-bacterioruberin and derivatives (e.g. mono-anhydrobacterioruberin and bis-anhydrobacterioruberin).^[41]

Methods for the identification of microorganisms in naturally occurring halites will likely be important for astrobiological studies during future Mars missions, including return samples. Since viable haloarchaea were isolated from terrestrian rock salt which is believed to be about 250 million years old, they can be considered long-term survivors^[42,43] and constitute a plausible analogue to potential halophilic life on Mars. In the present study, we simulated the embedding of various strains of haloarchaea into halite upon evaporation of hypersaline solutions and examined the samples by FT-Raman spectroscopy using laser excitation at 1064 nm and dispersive micro Raman spectroscopy at 514.5 nm.^[14]

The comparison of Fig. 6 with Fig. 4 clearly displays the difference in the spectra which one encounters when exciting resonantly^[14,53,55] (Fig. 6) and nonresonantly (Fig. 4).^[52]

In the non-resonant case, apart from the prominent carotenoid peaks due to the stretching modes of conjugated C=C and C–C bonds in the central chain, at 1507 cm⁻¹ and 1152 cm⁻¹, respectively, and the peak at 1002 cm⁻¹ being attributed to the in-plane rocking mode of CH₃ groups attached to the polyene chain, one obtains the signature from other cell components too,^[52] as nicely summarized, e.g. in Figs. 1B and 2B and in Table 1 of Ref. [20].

It is well known that any prokaryote (bacteria or archaea) consists mainly of protein (55% of its dry weight).^[56] In addition, prokaryotes contain approximately 23% nucleic acids, 9–10% lipids, 3% polysaccharides and smaller amounts of vitamins, metabolites and other compounds, which contribute to further peaks.

Peaks between 2800 and 3100 cm⁻¹, which belong to the CH₂ and CH₃ stretch and to the (C=C–H)_aromatic stretch, can therefore be mainly attributed to the presence of proteins^[20] (which are also signalled by the presence of the amide I band at around 1665 cm⁻¹, and in principle by the amide III band at around 1230–1295 cm⁻¹, here barely recognizable^[52]), and to a much lesser extent to the expected presence of characteristic core lipids of archaea (C₂₀ glycerol diphytanyl diether, or archaeol^[57]). The presence of the CH₂ and CH₃ stretch from proteins (and some lipids) automatically leads to the presence of the already mentioned peak due to the asymmetric CH₃ deformation and/or the CH₂ scissors vibration at around 1450–1460 cm⁻¹, which is also clearly recognizable. Note that the heights of the CH₂ and CH₃ peaks were used for the normalization of the spectra shown in Figs. ​Figs.44 and ​and55.

In the resonant case shown in Fig. 6, discrimination between carotenoids and other cellular constituents was not possible, since the spectral features associated with the stretching modes of conjugated C=C and C–C bonds in the polyene chain of the carotenoids and the in-plane rocking mode of CH₃ groups attached to the polyene chain get greatly enhanced, leading additionally to the appearance of characteristic overtones and combinations bands mainly in the spectral region between 2150 cm⁻¹ and 2650 cm⁻¹.^[55,62] Note therefore the comparatively low intensity of the peaks of the CH₂ and CH₃ stretch between 2800 and 3100 cm⁻¹.

In Fig. 7, the comparison of the spectrum of phenylalanine^[22] (an amino acid having a prominent peak at 1004 cm⁻¹ – compare again Figs. 1B, 2B, and Table 1 of Ref. [20] – and weaker peaks at 852 and 1602 cm⁻¹) with the spectrum obtained from the sample with H. dombrowskii H4 DSM14522^T embedded in halite demonstrates the detectability of this amino acid if its contribution to the overall spectrum is not overwhelmed by the presence of strong carotenoid peaks, as e.g. in the case shown in Fig. 7 of the H. salinarum strain DSM 3754.

Raman spectra of biological samples are complex and are still greatly underused as not all information which can be present^[22] can also be readily interpreted. Nevertheless, owing to the relative ease of acquisition of spectra, their reproducibility, together with the possibility of investigating water-containing samples, Raman spectroscopy has become a subject of increasing interest in many areas of biosciences,^[14,21] notably in the identification of microorganisms in environmental and other samples.^[59,60] Depending on the goal of the research work, one may prefer to rely on the information provided by the whole Raman spectrum, as delivered with non-resonant excitation of all the constituents of the sample under study, or one is interested in particular features of the Raman spectrum delivered by resonant excitation of well-selected constituents of the sample.^[18] Connected with each choice is a selection of advantages and disadvantages. In the case of non-resonant excitation in the NIR, the advantage is the avoidance of a disturbing fluorescence background, which can spoil the Raman signal in the case of strong fluorescence possibly appearing with resonant or near-resonant excitation in the visible (compare Fig. 4 with Fig. 6). On the other hand, one can gain in sensitivity when moving from the NIR to the visible due to the intensity of the Raman signal increasing with the 4th power of the wavenumber used for excitation,^[14] furthermore considerably enhancing the observed signal when reaching the conditions for resonance Raman scattering.

The results obtained in this study contribute to the database on Raman spectra of haloarchaea, which was suggested by Ref. [44]. Spectra from nine different strains of haloarchaea, representing four different genera, were obtained, clearly showing systematic differences in the intensity of the carotenoids peaks when the spectra are normalized against the protein (and lipids) peaks. The feasibility of using Raman spectroscopy with haloarchaea while the cells are embedded in the fluid inclusions of halite has been confirmed. Such crystals, on the inhospitable surface of Mars, could provide sufficient shielding from UV radiation that might allow the microorganisms within them to remain viable.^[38]

The non-resonant Raman spectra (Figs. ​(Figs.44 and ​and5)5) and the resonant Raman spectra (Fig. 6) of the haloarchaeal strains investigated in this study are dominated by characteristic carotenoid bands at 1507 cm⁻¹, 1152 cm⁻¹ and 1002 cm⁻¹;^[52,53] according to Ref. [53], they can be assigned to bacterioruberin, a C₅₀ carotenoid with 13 conjugated double bonds.^[41,61] These bands are situated at smaller wavenumbers than the corresponding bands in the C₄₀ carotenoids (containing 9–11 conjugated double bonds in the main chain),^[62,63] the C₄₀ carotene mix in Fig. 6 having peaks at 1520 cm⁻¹, 1157 cm⁻¹ and 1008 cm^−1[63] and the β-carotene in Fig. 7 peaks at 1515 cm⁻¹, 1157 cm⁻¹ and 1008 cm⁻¹.^[22,63]

The reason for this shift is well known, that is, the wavenumber position, e.g. of the C=C peak in a polyene chain being correlated with the number N of C=C bonds in this polyconjugated main chain^[64] (in other words with its length); thus the C=C peak position shifts from about 1585 cm⁻¹ for carotenoids with N = 4 (e.g. the C₂₀ carotenoid retinal) further down to 1515–1520 cm⁻¹ with N = 9 (e.g. the C₄₀ carotenoid α-carotene, β-carotene and lutein) and N = 11 (e.g. the C₄₀ carotenoid lycopene, with the peaks at 1510 cm⁻¹, 1157 cm⁻¹ and 1006 cm^−1[63]), and further down to about 1500 cm⁻¹ for carotenoids with N = 13 (e.g. in the case of the C₅₀ carotenoid bacterioruberin).^[52,53] The same behaviour holds for the other two characteristic carotenoid bands, although the shift is less pronounced. Small variations of the position for a given N are due to type and orientation of the terminal groups at the end of the polyene chain.^[62,63]

We did not attempt to extract and separate all possible carotenoids from our haloarchaeal samples for individual Raman spectroscopic observation, as it does not serve the purpose of developing a procedure to be used with extraterrestrial samples during planetary explorations (similar considerations being expressed already in Ref. [53]). Furthermore, in haloarchaea the structures of only some of the most important C₅₀ and C₄₀ ones have been identified in recent years.^[41] Nevertheless, besides the characteristic peaks attributable to C₅₀ carotenoids of haloarchaea, the slight asymmetry of these peaks towards higher wavenumbers might indicate the presence of C₄₀ carotenoids in small amounts, lycopene and β-carotene being probably the most abundant.

Apart from the fact that, as Figs. ​Figs.44 and ​and55 show, there are significant differences in the concentration of carotenoids in different haloarchaeal species,^[65] variations in the intensity of the pigment bands across the size of colonies of haloarchaea can occur due to cell motility and cell concentration;^[52] additionally, the carotenoid content of some haloarchaea has been found to be related also to their growth conditions.^[66,67] As stated in the experimental section, we performed our Raman spectroscopic measurements after growth to late logarithmic phase or early stationary phase.

Additional considerations about biomarkers for the detection of haloarchaea^[53,61] need to be further elucidated; for example the well-known membrane protein bacteriorhodopsin has been suggested by Ref. [53] as a second biomarker for haloarchaea in hypersaline environments. Bacteriorhodopsin (which is probably the simplest proton pump known^[68]) contains retinal, a C₂₀ chromophore with four conjugated C=C bonds in the main chain; Fig. 6 of Ref. [53] shows the differences in the resonance Raman spectra of isolated bacteriorhodopsin (purchased as a commercial preparation, the C=C peak situated at 1536 cm⁻¹) and of bacterioruberin (from cells of H. salinarum). More details on resonance Raman spectra of bacteriorhodopsin can also be found in Ref. [69]. Interestingly, all the carotenoid peaks shown in Figs. ​Figs.44–6 of the present study are quite consistent with the resonance Raman spectrum of bacterioruberin shown in Fig. 6 of Ref. [53], giving no significant evidence for peaks attributable to the presence of bacteriorhodopsin. Indeed, bacteriorhodopsin was produced by H. salinarum NRC-1 and Halorubrum sp. strain NaxosII, under the growth conditions used here, with about 1–2% of total membrane proteins, as we prepared it from haloarchaeal purple membranes, containing bacteriorhodopsin,^[70] as described in Ref. [71]. Polyacrylamide gel electrophoresis was performed according to Laemmli,^[72] and showed a protein band of molecular mass of 26 000 Da. Yet, any characteristic bacteriorhodopsin peaks were likely obscured by the overwhelming presence of the characteristic carotenoids peaks. Additionally, bacteriorhodopsin is not produced by all haloarchaea; its gene was found to be absent from some genera such as Halococcus, Haloferax and others;^[23,65] in addition, its expression can vary significantly, depending on environmental conditions.^[73] The spectra in Figs. ​Figs.44–6 imply therefore that, in contrast to the proposal by Ref. [53], bacteriorhodopsin cannot be considered an ideal marker for the detection of haloarchaea.