The durability of building materials, including those of historical value, is influenced by numerous physical, chemical, and biological factors. Among them, salt crystallization, which forms as a result of salt migration with capillary water and subsequent drying out and precipitation, poses a serious threat to the materials' structure (Stryszewska, 2014). Moreover, salt-attacked monuments constitute a suitable habitat for halophilic microorganisms (Piñar et al., 2014; Adamiak et al., 2015; Otlewska et al., 2017) as these can easily adapt to salty micro-niches available on a building's surface, thus contributing to its structural destabilization (Saiz-Jimenez and Laiz, 2000; Piñar et al., 2009; Adamiak et al., 2015, 2016).

Microbial metabolomics has proved to be an emerging field for microorganism identification, mutant screening, and functional gene research, metabolic pathway identification, and microbial engineering. Over the last two decades it has provided new insights on the activities of microbes in response to environmental conditions (Xu et al., 2014; Villas-Boas, 2016). Recently, several outstanding papers on microbial metabolomics have been published (Rochfort, 2005; Mashego et al., 2007; van der Werf et al., 2007; Tang, 2011; Baidoo et al., 2012; Xu et al., 2014; Barkal et al., 2015). To our knowledge, there are only a limited number of reports on biodeterioration and biocorrosion processes (Lenhart et al., 2014; Brauer et al., 2015; Gutarowska et al., 2015; Bonifay et al., 2017; Szulc et al., 2017), and none of them has focused on halophilic microorganisms.

Biodeterioration of building materials due to active growth of microorganisms is another area where microbial metabolomics has much to offer. For example, Gutarowska et al. (2015) used ultra-high performance liquid chromatography coupled to high-resolution mass spectrometry (UPLC/HRMS) to investigate activated pathways of microbial communities inhabiting wood and brick collected from historical objects. Szulc et al. (2017) compared the metabolite profiles of molds on carton-gypsum board using a high-resolution surface-assisted laser desorption/ionization time-of-flight mass spectrometry based on a gold nanoparticle-enhanced target (AuNPET SALDI-ToF-MS) imaging method. Growing awareness of biodeterioration of building materials and halophiles contributing to this phenomenon (Rölleke et al., 1998; Heyrman et al., 1999; Laiz et al., 2000, 2009; Piñar et al., 2001, 2014; Ripka et al., 2006; Ettenauer et al., 2014) has highlighted gaps in our understanding of the metabolic pathways that characterize halophilic communities colonizing historical buildings. At present, there are no studies providing an in-depth metabolomic assessment of halophiles-induced biodeterioration.

A well-known phenomenon found on wall surfaces is the formation of salt efflorescence, which mimics the extreme salty conditions favoring the proliferation of halophiles (Sterflinger and Piñar, 2013). The ability of halophilic microorganisms to cope with osmotic stress in hypersaline environments is an important adaptation. Two different strategies for coping with osmotic stress have been reported. One involves the accumulation of inorganic ions (mostly K⁺ and Cl⁻ rather than Na⁺) and is known as the “salt-in-cytoplasm” strategy. The other, known as the “compatible solute” strategy employs the uptake or synthesis of organic molecules, namely sugars and polyols or α- and β-amino acids with their derivatives (Madigan and Oren, 1999; Xiang et al., 2008; Averhoff and Müller, 2010). When studying the interactions of halophilic microorganisms with building materials, we should take into consideration not only their ability to survive in a high-salt environment (haloadaptation), but also their contribution to the degradation of building materials. A wide range of carotenoid pigments, such as β-carotene, α-bacterioruberin and derivatives, and salinixanthin is produced by halophilic microorganisms (Oren, 2002). They are the major cause of typical rosy stains, significantly influencing the visual appearance of building material (Sterflinger and Piñar, 2013). Organic acids also play an important role, because they contribute to cracking and detachment of building materials (Gutarowska and Czyzowska, 2009). Halophilic microorganisms isolated from salterns, brines, sediments, and soil were shown to produce organic acids; e.g., gluconic acid, 2-ketogluconic, citric acid, formic acid, succinic acid, fumaric acid, and acetic acid (Tomlinson and Hochstein, 1972; Vreeland et al., 1998; Yadav et al., 2015); however, this ability was not confirmed for isolates from a building environment.

When elucidating the biodeterioration caused by halophilic microorganisms, it can be helpful to utilize microscopic methods. However, rapid detection of microbes in their natural environment (Hong et al., 2016) is still a challenge. Here, we report the use of coherent anti-Stokes Raman scattering (CARS) for in situ detection of a single living cell. This state-of-the-art technology offers a non-invasive, label-free, and direct way to image biological samples with chemical sensitivity (Cheng et al., 2002; Cheng and Xie, 2004; Rodriguez et al., 2006). It relies on the abundance of CH₂ groups in lipids and the distinctive CH₂ stretch vibration frequency at 2,840 cm⁻¹ (Folick et al., 2011), thus enabling detection of a single bacterium. CARS requires the input of two laser pulses, a pump-probe beam and a Stokes beam, which interact with the specimen to generate an anti-Stokes field. When the frequency difference between the pump and the Stokes pulses matches the frequency of a certain molecular vibration, a strong signal is generated, allowing molecule-specific imaging (Zhang et al., 2015; Hong et al., 2016). Visualizing lipid structures in live cells may therefore represent an alternative to standard procedures employed in biodeterioration studies.

The present work focused on untargeted metabolomics using high-performance liquid chromatography coupled to high-resolution mass spectrometry (HPLC/HRMS) to explore the halophile metabolome. Mass spectrometry is known to be the most sensitive and selective technique among many platforms. This work demonstrates that a combination of what might be termed “community metabolomics” and state-of-the-art CARS microscopy has an enormous potential for the comprehensive characterization of biodeterioration on salt-attacked buildings. Therefore, the aim of the research was to: (1) identify groups of metabolites produced by mixed cultures of halophilic microorganisms that may be potentially responsible for degradation of brick; (2) reveal the pattern of metabolic responses of halophilic microorganisms to severe saline conditions; and (3) visualize live cells of halophilic microorganisms with CARS.

The main objective of the present study was to identify the pattern of metabolic responses by halophilic microorganisms exposed to different salts present in brick samples. For this purpose, we sought to isolate groups of metabolites potentially responsible for the degradation of building materials, as well as those that play a vital role in haloadaptation to severe saline conditions.

The metabolome was found to be largest in the sample saturated with MgSO₄ (838 compounds) and smallest in the sample saturated with KCl solution (574 compounds). The numbers of putative identifications were noticeably different between samples (Table 2). Similarities and differences between the samples from different backgrounds separated into distinct clusters, as defined by principal component analysis (PCA) plots (Figure 2). According to the PCA model, total variance of the data was 55.3%, with 34.8% and 20.5% attributed to PC1 and PC2, respectively, and an additional 19.4% contributed by PC3 (10.3%) and PC4 (9.1%). PC1 and PC2 results (Figure 2A) show clear clustering of inoculated and uninoculated samples. The samples saturated with different salt solutions and inoculated with mixed cultures of halophilic microorganisms are spread across two different clusters. The first cluster (red dashed line) is dominated by the inoculated brick samples exposed to the corrosive effect of KCl, NaCl (1st cycle), and MgSO₄. The second cluster (orange dashed line) contains exclusively the inoculated samples saturated with Mg(NO₃)₂ and NaCl (3rd cycle). By comparison, results defined by PC3 and PC4 (Figure 2B) show structured responses to different salt types. Chloride salts (blue dashed line) and magnesium salts (navy dashed line) are clustered separately, as are samples exposed to 3 cycles of NaCl (purple dashed line), which also form a distinct group. Such clustering may be explained by previous findings, whereby halophiles isolated from historical artifacts showed higher tolerance to magnesium salts (Oren, 1994; Saiz-Jimenez and Laiz, 2000). This feature may play an important role in haloadaptation (De Medicis, 1986). In contrast, a minimum concentration of Na⁺ ions is sufficient for growth, due to their impact on transport processes in the cell membrane (Ventosa et al., 1998). According to the literature, Mg²⁺ ions are required for activation of biosynthetic systems, whereas KCl and NaCl have a mostly inhibitory effect (Stern and Tietz, 1978; Oren, 2002). Here, PCA plots were constructed based on an unsupervised method and thus only a qualitative separation is showed. PCA allowed us to detect sample patterns and the differences between them. As the latter remained somewhat unclear, we used partial least squares discriminant analysis (PLS-DA) to achieve a better separation between samples and evaluate the causative factors for the observed clustering (Figure 2C). The PLS-DA plot based on putatively identified metabolites, revealed a clear separation between uninoculated (dark green dashed line) and inoculated samples (light green dashed line). Accordingly, the abundances (concentrations) of putative metabolites may be related to the microorganisms themselves and their ability to produce metabolites potentially associated with brick deterioration, rather than a specific type of salt. Figure 2 presents strong evidence for the existence of clear metabolic differentiation as a consequence of halophile response to a saline environment.

Metabolomic data indicated the presence of intermediates presumably involved in 26 metabolic pathways related to both degradation of building materials and adaptation to severe saline environment (Table 3). The identified pathways included: biosynthesis and degradation of amino acids (e.g., valine, lysine, isoleucine), sulfur metabolism and carbohydrate digestion, carotenoid biosynthesis (salinixanthin), mineral absorption via sodium and chloride transporters, and glycerophospholipids metabolism, which plays a crucial role in the formation of biological membranes. A complete list of pathways and the detected intermediates is presented in Supplementary Materials 3.

Metabolomics studies have provided answers to numerous basic questions regarding the adaptation of halophiles to salty environments and their ability to degrade building materials. In our study, we found metabolites potentially involved as intermediates in metabolic pathways such as synthesis of compatible solutes (betaine). This finding supports the mechanism allowing for osmoadaptation of halophilic microorganisms found in building materials. According to the “salt-in-cytoplasm” strategy, microorganisms adapt the protein machinery of the cell to augment the salt concentration in the cytoplasm. In contrast, identification of betaine synthesis pathways supports the action of the organic-osmolyte strategy, which is based on the uptake or synthesis of organic molecules so that their concentration within the cell is regulated according to the salt concentration outside the cell. The ability of aerobic heterotrophic Bacteria to synthesize betaine has not been proved yet; however, betaine synthesis was reported in phototrophic Bacteria and Archaea via the serine pathway with choline as an intermediate (Kunte et al., 2002). Besides playing a crucial role in haloadaptation, compatible solutes can stabilize proteins or even whole cells, protecting them against heat, desiccation, freezing and thawing, and denaturants such as urea (Lippert and Galinski, 1992; Kunte et al., 2002). Interestingly, it was found that members of the genus Halobacillus, which are mainly chloride-dependent bacteria, were able to switch their osmolyte strategy as a consequence of changing salinity (Ma et al., 2010). Moreover, our results showed also the biosynthesis of a ketocarotenoid similar to salinixanthin. One of the possible ways for its formation is synthesis from β-carotene. The formation of ketolated carotenoids from β-carotene is rather rare and was observed for the first time for marine bacteria (Ralley et al., 2004). UPLC coupled to diode-array detector electrospray ionization mass spectrometry (UHPLC-DAD-ESI-MS) of inoculated brick samples confirmed, however, the presence of β-carotene (Supplementary Materials 4), thus supporting the above-mentioned theory. β-carotene was present in every inoculated brick sample, but was most abundant in the sample saturated with KCl solution. This finding emphasizes the destructive potential of halophilic microorganisms, in particular their ability to cause aesthetic changes on the surface of buildings. The changes in color of inoculated brick samples immersed in corrosive salt solutions, were corroborated by spectrophotometric measurements using the CIEL^*a^*b^* system (Table 4). Thus, inoculation with halophilic microorganisms and yearlong incubation at 28°C and 80% relative humidity resulted in a statistically significant increase in brightness (L^*), a decrease in redness (a^*) and yellowness (b^*), as well as a total color change exceeding a value of 5 on the surface of samples. Generally speaking, salty deposits mimic extreme saline habitats, hostile to halophilic microorganisms and contribute to further contamination of building materials, which may be observed as yellowish to reddish discolorations caused by production of carotenoids. Primarily, pigment formation is associated with a response against photooxidative damage, chemical, and/or salt stress. On salt-attacked objects, biogenic pigments are usually very resistant, even after the microorganisms' death (Piñar et al., 2014). Previous studies held by Ettenauer et al. (2014) and Piñar et al. (2014) on objects of art, revealed that microbial metabolites, especially pigments, altered building elements, and were thus major contributors to biodeterioration. Here, the biocorrosive potential of halophilic microorganisms was also proved by the formation of trace amounts of acids, in particular acetic acid (Supplementary Materials 5). Organic acids are known to be major causes of biodeterioration of building materials (Gutarowska and Czyzowska, 2009).

Among putative metabolites, lipids were one of the dominant compound groups. They provide information pertaining to an organism's actual phenotype, variety of structural and functional properties, cellular architecture, and function (Harkewicz and Dennis, 2011). Among lipids, phospholipids (PLs) were the most abundant category, particularly neutrally charged phosphatidylethanolamine (PE) and phosphatidylcholine (PC), and negatively charged phosphatidylserine (PS), phosphatidylinositol (PI), and phosphatidylglycerol (PG). As expected, hierarchical clustering (Figure 3) showed a different lipid pattern in the sample exposed to higher salinity (sample II_3). Exposure to 3 cycles of a NaCl corrosive environment induced a decrease in the concentration of PS, while only slightly reducing the concentration of PE. A specific PL composition may be essential for haloadaptation (Oren, 2002). Any change in lipid composition resulting from exposure to a saline environment is due to preservation of the membrane's bilayer structure. To preserve its integrity, a correct proportion of bilayer-forming and non-bilayer-forming PLs is required. Generally, PE tends to form non-bilayer phases, whereas PG forms a stable bilayer. In this respect, increasing salinity results in an increase of negatively charged PLs at the expense of PE. Accordingly, the slightly lower PE concentration observed in our study when salinity was augmented, fits in part with this pattern, but does not allow us to draw any general conclusions. Interestingly enough, we noted a remarkable change in membrane composition due to a decrease in PS levels. This shift may be explained by immediate conversion to PE, of which PS is a precursor (Vance and Steenbergen, 2005). Furthermore, an increase in medium salinity induced an increase in the concentration of unsaturated fatty acids in the membrane. The major fatty acids detected in the sample with higher NaCl concentration (II_3) were C24:2^{Δ5, 13} and C20:1^Δ13. Changes in fatty acid composition represent an important factor in the maintenance of membrane fluidity (Turk et al., 2004). Unsaturated fatty acids generally increase membrane fluidity (Oren, 2002).

Finally, we applied CARS microscopy to elucidate biodeterioration and detect single halophilic microorganisms involved in this process in situ. Figures 4A–E shows a 2,850 cm⁻¹ CARS image of the mixture of halophilic microorganisms in building material (red points in the picture). The corresponding 3D image (Figure 4F) provides a full picture of halophiles' distribution across the sample. Specifically, we examined the influence of environmental conditions (saturation in different salt solutions) on the abundance of halophilic microorganisms in building material. It is noteworthy, that in samples saturated with magnesium salts (Figures 4A,B) halophilic microorganisms are more abundant than in samples saturated with chloride salts (Figures 4C–E), which may be due to the aforementioned inhibitory effect of KCl and NaCl. The average diameter of the cell was 0.80 ± 0.08 μm. According to the 3D image (Figure 4F), halophiles were present across the entire sample (up to 25 μm). Although, the present study focused on halophiles detection, their metabolic activity may be studied by monitoring spectral changes (Hellerer et al., 2007) and thus provide clues of how haloadaptation occurs. In particular, viability of halophiles entrapped in small inclusions, following drying and the formation of crystals on the surface of building material, is a key topic in haloadaptation (Ma et al., 2010).

The main focus of the present study was to find a relationship between metabolomic data and activity of halophilic microorganisms in an attempt to better understand how they contribute to biodeterioration of building materials or cope with osmotic stress. Our results clearly demonstrate the usefulness of halophilic community metabolomics. The information produced by metabolomics and its very close link to the phenotype of microorganisms makes it possible to successfully find biomarkers, which could be used for early detection of biodeterioration. In the case of halophiles, carotenoids represent possible candidates. Although we found several metabolites, many of them remained unidentified. Therefore, a reliable evaluation of biodeterioration caused by halophilic microorganisms will necessarily have to extend beyond already identified metabolites. Of particular relevance for biodeterioration studies is to combine various metabolomics methods, separation modes, and platforms to significantly improve data integration on the role played by halophilic microorganisms. This may also reveal other mechanisms of action and complex interactions in a salty environment, as the one encountered in building materials. Until now, it was challenging to rapidly detect microbes in their natural environment; however, CARS microscopy offers unique advantages in this sense. Indeed, CARS may soon become an indispensable tool for lipid-related in situ detection of a single living cell with high sensitivity and selectivity, without alterations to the sample. In the future, it may become a crucial instrument for assessing biodeterioration of cultural property caused by halophilic microorganisms, where the amount of a sample is significantly limited.

JA was responsible for MS analyses, data analysis, CARS microscopy analysis and took the lead in writing the manuscript. VB conducted MS analyses and assisted with data analysis and writing. AO took part in sample preparation, in particular preselection of strains and inoculation, assisted with CARS microscopy and manuscript proofreading. JS guided the MS analyses and contributed to the results proofreading. IB contributed to the MS analysis and results proofreading. TS was responsible for the brick sample preparation, especially saturation with salty corrosive solutions. SK was involved in collecting the brick samples and their further preparation. JO was involved in β-carotene identification, optical measurements and manuscript proofreading. DŻ contributed to the determination of color parameters and manuscript proofreading. BG contributed to the ideas, writing and editing the manuscript.