Introduction

Stone cultural heritage suffers persistent deterioration from multifactorial stressors during prolonged outdoor exposure1. Its structural instability results from the complex synergy of physical, chemical, and biochemical processes. These include physical property changes and embrittlement caused by temperature/humidity variations2,3, acidification from atmospheric pollution and acid rain4, and biological erosion due to microbial colonization5,6. Microbial deterioration is a pervasive and inevitable phenomenon, primarily involving physical disruption and chemical degradation. Physical disruption typically manifests as mechanical stress from microbial growth, where fungal hyphae or microbial cells expanding within stone pores and fractures exert pressure that compromises structural integrity. Chemical degradation begins with biofilm colonization by diverse communities (including bacteria, fungi, archaea, algae, and lichens)7,8,9,10,11,12. Their metabolic activities, particularly in carbon (C), nitrogen (N), and sulfur (S) cycling, damage stone through acid production, salt crystallization pressure, and mineral redox transformations6,9,13,14,15. Microbially induced acid attack poses an insidious and persistent threat to cultural heritage. Research on limestone sculptures at the Longmen Grottoes has identified 21 bacterial strains with significant biodegradative capabilities (i.e., acidophilic, methyl-oxidizing, and sulfur-oxidizing). The study also revealed that these bacteria deteriorate stone by secreting organic or sulfuric acids, which lower the local pH and dissolve calcium and magnesium carbonates, thereby establishing a valuable strain database for conservation science16.

A surge has been witnessed in research on microbial deterioration of stone cultural heritage in recent years. The study of microbial deterioration of stone cultural heritage has undergone a significant evolution, shifting from descriptive cataloging of surface microbial communities to a functional investigation of metabolic mechanisms, largely driven by the application of multi-omics technologies17. A central focus of current research is the understanding of coupled microbially-driven nitrogen and sulfur cycles, which synergistically accelerate decay through acid production and salt damage18,19,20,21. Research by He et al. on the West Lake UNESCO World Heritage Site demonstrates that nitrifying bacteria exhibit broad ecological niches, potentially participating in the initiation, succession, and expansion of stone biodeterioration22. Research on the Kaiping Diaolou directly links anthropogenic nitrogen enrichment to the dominance of acid-producing ammonia-oxidizing bacteria23, while long-term research at Angkor Wat employs stable isotope tracing to reveal a complex internal nitrogen cycling14,24. Research integrating DNA/RNA-targeted sequencing and functional gene analysis indicates that the microbial communities on Portchester Castle sandstone harbor extensive metabolic diversity, enabling near-complete nitrogen/sulfur cycling, autotrophic carbon fixation, and mineral transformation, highlighting their key biogeochemical roles in stone heritage ecosystems25. Furthermore, metagenomic and metaproteomic analyses of sites like the North Grottoes Temple provide direct evidence of highly active microbial metabolism centered on nitrogen and sulfur transformations, moving beyond genetic potential to demonstrate active deterioration processes17. Collectively, these studies underscore that future conservation strategies must target these key metabolic pathways.

Biogeochemical cycles in natural ecosystems function as interconnected networks driven by the synergy of biotic and abiotic factors, along with their dynamic feedback26,27,28. Nitrogen and sulfur cycles frequently couple through electron donors and acceptors, particularly sulfur-driven autotrophic denitrification. These microorganisms simultaneously perform sulfur oxidation and denitrification, removing NO3⁻ while accumulating SO42⁻29,30. This process can be accomplished either by single autotrophic microorganisms (e.g., α-, β-, γ-, and ε-Proteobacteria) or by cooperation among those microorganisms31,32,33, revealing multiple potential synergistic mechanisms between denitrification and sulfur oxidation34. The sulfate generated during sulfur oxidation exhibits high reactivity and readily precipitates with Ca2+ within the stone to form gypsum (CaSO4·2H2O). The volumetric expansion accompanying gypsum crystallization exerts crystallization pressure, forcing open rock pores and ultimately leading to surface spalling and irreversible damage35. Despite its significance, the coupling of nitrogen and sulfur cycles on stone cultural heritage and its resultant deterioration impacts are not yet fully understood. To this end, this study employed high-throughput sequencing, qPCR, DNA-stable isotope probing (DNA-SIP), and IRMS to analyze the specific microbial communities, nitrogen cycling processes, and nitrogen-sulfur coupling within the stone materials of the Former Residence of Chen Cihong. Crucially, this work provides novel insights into the acid attack mechanisms of stone cultural heritage and informs future conservation strategies.

Methods

Sample collection and detection of physicochemical properties

The Former Residence of Chen Cihong (Fig. S1), located in Shantou City, Guangdong Province, China, is composed primarily of heritage mansions, such as Shànjū Shì (built 1922), Lángzhōng Dì (1910), and Shòukāng Lǐ (1922). This architectural complex synthesizes Chinese Lingnan architectural elements with strong Western features and influence. Its primary structures, along with interior wood carvings and stone carving reliefs, embody traditional Chinese forms. Walls are constructed of brick masonry, while balconies and corridor bridges employ granite as the primary building material. Sampling was exclusively conducted on outdoor surfaces of Shànjū Shì with visually dense biofilm coverage. Differential exposure regimes to sunlight and precipitation were observed across floors: the third floor features a balcony fully exposed to weather elements (with restricted tourist access), while the first and second floors are sheltered. Biofilm samples were collected from diverse building materials using non-invasive sampling tools (sterile swabs and transparent adhesive tapes). Sampling included: 1) stone materials: 1st floor (S1), 3rd floor (S3); 2) tiles: 2nd floor (C2), 3rd floor (C3); 3) mosaic glass: 3rd floor (G3); 4) wood and wooden boards: 1st floor (B1, W1) (Fig. 1a). All samples were labeled, sealed, and transported at 4 °C to the laboratory.

Fig. 1: Biofilm samples collected from different materials in the Former Residence of Chen Cihong and a stable isotope-labeled cultivation workflow.
Fig. 1: Biofilm samples collected from different materials in the Former Residence of Chen Cihong and a stable isotope-labeled cultivation workflow.
Full size image

A total of 7 biofilm samples were collected (a). Among these, sample S3 from stone substrates was isotopically labeled using 14NH4Cl and 15NH4Cl as nitrogen sources; sampling occurred at Days 1, 3, and 7. Microbial DNA extracted from the biofilms underwent DNA-SIP, while the culture supernatants were subjected to IRMS analysis (b).

Pre-weighed biofilm samples from different materials were desiccated in an oven until constant weight was achieved. The post-desiccation mass was recorded to calculate water content. Subsequently, aliquots of pre-weighed biofilm samples were vortex-mixed in ultrapure water for pH determination using a calibrated pH meter. Aliquots (1 mL) of this suspension were subjected to 10-fold dilution, then filtered through 0.22 μm sterile membranes. The filtrate was analyzed for anions (NO3⁻, SO42⁻, etc.) immediately via ion chromatography (925 CN, Metrohm).

DNA-SIP cultivation and gene abundance quantification

Fresh stone biofilm from S3 was selected and pre-incubated in the darkness at room temperature for 24 h to deplete nitrogen reserves. Subsequently, 1 g of the pre-treated biofilm was weighed and transferred to 100 mL of N-free mineral salt medium (MSM). The MSM contained (per liter): 0.8 g Na2HPO4, 0.2 g KH2PO4, 0.2 g MgSO4, 10 g NaCl, and 1 mL of trace element stock solution, dissolved in deionized water, with the pH adjusted to 7.0. This medium was amended with either 14NH4Cl or 15NH4Cl (Sigma-Aldrich, >98% atom) as the nitrogen source, each at a final concentration of 1 mM, each treatment with two replicates. Sealed cultures were incubated and then collected on days 1, 3, and 7 (Fig. 1b). Genomic DNA was subsequently extracted from the collected samples using the CTAB-based method described by Zhou et al.36.

Genes encoding key enzymes involved in nitrogen transformation processes were amplified by PCR. The amplicons were ligated into the pMD19-T vector (Takara) and sequenced to generate recombinant plasmids harboring the respective gene fragments. Detection of the cycle threshold (Ct) values for the target genes was performed using the PerfectStart® Green qPCR SuperMix kit (TransGen Biotech) via quantitative real-time PCR (qPCR). Reactions were conducted in triplicate with a reaction volume of 10 µL on a LightCycler® 480 II Real-Time PCR System (Roche). The amplification primers and reaction programs are listed in Table S1. To construct the standard curve, newly extracted recombinant plasmids were serially diluted 10-fold (10⁻1 to 10⁻9). Based on this standard curve, the absolute copy numbers of the target genes were calculated to quantify gene abundance.

SIP gradient fractionation

DNA extracted from incubations (Day 3) was subjected to isopycnic density gradient centrifugation as modified from Zhang et al.37,38. Approximately 3000 ng of 15N-DNA was added to Cesium chloride (CsCl) density gradient solution (1.85 g/mL, prepared by dissolving 50 g CsCl in 30 mL GB buffer [0.1 M Tris-HCl, 0.1 M KCl, 1.0 mM EDTA]). The refractive index (RI) was adjusted to 1.4029 ± 0.0002 using a calibrated refractometer (ATAGO PAL-RI, Japan). The mixture was transferred into 6 mL ultracentrifuge tubes (Beckman Coulter; Cat. No. 344058). After heat-sealing and balancing, tubes were symmetrically loaded into a Beckman Coulter 32 Ti fixed-angle rotor (Cat. No. 362754) and ultra-centrifuged (Beckman Coulter OPTIMA L-100K) under gradient separation conditions (20 °C, 190,000 × g, 44 h; without break). After centrifugation, fractions were displaced from the bottom to the top by elution with Milli-Q water using a NE-1000 fixed-flow rate pump (New Era Pump Systems, Farmingdale, NY, USA) at 0.34 mL/min. One fraction per minute was collected, yielding 15 density gradient fractions per sample. The refractive index (RI) of all fractions was measured (Table S2). Buoyant density (BD) was calculated from RI values using the equation: BD = (RI × 10.8601) – 13.4974.

DNA purification and qPCR verification

To the fractionated genomic DNA solution from density gradient ultracentrifugation, 550 μL of Polyethylene Glycol 6000 solution was added (prepared by dissolving 150 g PEG 6000 and 46.8 g NaCl in deionized water, brought to a final volume of 500 mL, followed by filtration through a 0.22 μm membrane filter and sterilization). The mixture was incubated statically at room temperature for 2 h to precipitate DNA. Subsequent centrifugation at 13,000 × g for 30 min was followed by washing with 70% ethanol and drying. The purified fractionated genomic 14N-DNA and 15N-DNA were then dissolved in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and stored at – 80 °C.

The amoA gene was amplified using primers CA377f/C576r39. Following the method described in section “SIP gradient fractionation”, qPCR was performed using fractionated genomic 14N-DNA and 15N-DNA as templates to quantify the abundance of the target amoA gene. The thermal cycling conditions were: 50 °C for 2 min, 95 °C for 10 min; 40 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s. A standard curve was generated by serial 10-fold dilutions (10⁻1 to 10⁻9) of newly extracted recombinant plasmids. The copy numbers of the target gene were calculated based on this standard curve, and gradient curves illustrating the labeling efficiency were plotted according to qPCR results.

IRMS and high-throughput sequencing

The supernatant from DNA-SIP was filtered through 0.22 μm membrane filter. The analytes (NH4+, NO2⁻, and NO3⁻) were combusted to N2, and the relative isotopic abundance (Rsq) was measured using an isotope ratio mass spectrometer (Thermo Fisher MAT253). The nitrogen isotopic ratio δ 15N (‰) in gas samples was then calculated against atmospheric N2 (Rst) as the standard40.

Genomic DNA from biofilm of different materials was extracted. Following quality and purity assessment, the DNA was sent to MAGIGENE Technology Co., Ltd. (Guangzhou, Guangdong, China) for amplicon sequencing according to standard protocols using the Illumina MiSeq PE250 platform, targeting the bacterial V4-V5 segment (forward primer: GTGCCAGCMGCCGCGGTAA; reverse primer: CCGTCAATTCMTTTRAGTTT), and the fungal ITS1 region (forward primer: GGAAGTAAAAGTCGTAACAAGG; reverse primer: GCTGCGTTCTTCATCGATGC). Three replicates were prepared. For fractions DNA, samples corresponding to peak gradients in qPCR validation curves and adjacent fractions (representing heavy 15N-DNA fractions) were selected for metagenomic sequencing (designated 3-9, 3-10, 3-11, and 3-12) (Fig. 3c). The raw sequencing data were deposited in the NCBI database with accession number PRJNA1314663, PRJNA1314667, and PRJNA1311989.

Bioinformatics analysis

Raw sequencing data for bacterial and fungal diversity were processed using the dada2 module in QIIME2 v2023.541. Operational Taxonomic Units (OTUs) were annotated for bacteria using the SILVA database (https://www.arb-silva.de/) and for fungi using the UNITE database (https://unite.ut.ee/). Subsequent diversity analyses were performed with the vegan package in R (https://cran.r-project.org/web/packages/vegan/index.html).

Raw reads of metagenomic sequencing were quality controlled by fastp software42, and clean reads were assembled by MEGAHIT v1.2.9 (–k-min 19 –k-max 149 –k-step 20)43. Gene prediction and functional annotation of the assembled contigs were performed using Prodigal v2.6.344 and EggNOG-mapper v2.0.145, respectively. Contigs were subsequently aligned and binned with BWA v0.7.1746, samtools47, and MetaBAT248. The resulting bins were evaluated with CheckM v1.1.349, and only metagenome-assembled genomes (MAGs) with completeness larger than 70% and contamination smaller than 10% were reserved. Specifically, functional annotation of genes involved in nitrogen and sulfur metabolism was conducted on both prodigal-predicted contigs and binned genes. Data visualization was implemented via the online platform ChiPlot (https://www.chiplot.online/).

Results

Physicochemical properties and microbial community on different materials

To characterize abiotic factors influencing biodeterioration, NO3⁻, SO42⁻, Cl⁻, pH, and moisture content were measured across biofilm samples of different materials (Table 1). The content of NO3⁻ was higher in samples S1 and W1 than that in others, whereas the contents of SO42⁻ were higher in S3 and B1. In addition, the highest content of Cl⁻ occurred in B1. Samples S3 and G3 exhibited slight acidity, while S1 and W1 showed slight alkalinity. Moisture content was highest in sample S1, intermediate in W1 and S3, and lowest in G3.

Table 1 Physicochemical properties of different materials in the Former Residence of Chen Cihong
Full size table

Biofilm communities on different materials were evaluated by high-throughput sequencing of bacterial 16S rRNA gene and fungal ITS1 region. Bacterial OTUs were classified into 33 phyla, 1,213 genera, while fungal OTUs comprised 6 phyla, 55 genera. Among bacteria, Proteobacteria and Cyanobacteria predominated in microbial communities of all the samples at the phylum level (Fig. S2). For fungi, unclassified OTUs and Ascomycota showed higher abundance proportionally (Fig. S3). Diversity analysis revealed the highest bacterial diversity in S3, followed by S1, B1, and W1, with G3 the lowest. Conversely, the fungal diversity was highest in C2 and G3, followed by S1, while B1 and W1 showed the lowest values (Fig. S4).

Notably, 10 genera involved in ammonia oxidation, nitrification, and anaerobic ammonium oxidation (anammox) were detected across different samples (Fig. 2). These included AOA (Candidatus Nitrocosmicus, Nitrososphaeraceae, and Candidatus Nitrososphaera), AOB (uncultured Nitrosomonadales bacterium, Nitrosomonas, and Nitrosomonadaceae), NOB/Comammox (Nitrospira and Nitrosospira), and anaerobic ammonium-oxidizing bacteria (anammox bacteria, uncultured and enriched clone Anammox_49). Among these, Nitrospira was present in all samples. Furthermore, 8 genera were detected in S3, which exhibited the highest abundances of Candidatus Nitrocosmicus and Nitrospira. Sample S1 also showed a relatively high abundance of Nitrospira (Table S3). Collectively, stone materials (specifically S3) in the Former Residence of Chen Cihong harbored nitrogen-cycling microorganisms with higher diversity and abundance.

Fig. 2: Taxonomic composition of ammonia-oxidizing, nitrifying, and anammox bacteria in biofilms on building materials from the Former Residence of Chen Cihong.
Fig. 2: Taxonomic composition of ammonia-oxidizing, nitrifying, and anammox bacteria in biofilms on building materials from the Former Residence of Chen Cihong.
Full size image

Taxonomic distribution of relevant genera derived from SILVA-annotated OTUs, with different colors indicating distinct taxonomic groups.

Increasing abundance of genes driving key nitrogen transformation reactions

Functional genes involved in nitrogen transformation were quantified in 14N-DNA and 15N-DNA using DNA-stable isotope labeling. The copy numbers of amoA, nxrB, norB, and nosZ genes increased over time of incubation (Fig. 3a, b). In the 15N-treated group, the copy number of amoA gene (encoding ammonia monooxygenase) increased from 5.25 × 102 on day 1 to 6.22 × 103 on day 7, while nxrB gene (encoding nitrite oxidoreductase) increased from 3.83 × 103 to 5.37 × 106. Additionally, norB gene (encoding nitric oxide reductase) increased from 2.51 × 103 to 1.15 × 105, and nosZ gene (encoding nitrous oxide reductase) increased from 1.67 × 105 to 2.69 × 107 (Table S4; Fig. S5). The higher abundance and greater increase in denitrification genes (norB and nosZ) suggest that microbial biofilm harbors not only active ammonia oxidation and nitrification reactions, but also more active denitrification on the stone material of the Former Residence of Chen Cihong. IRMS analysis revealed a decreasing trend in δ (‰) of 15NH4+ in the labeled culture medium over time (Fig. 3d). Concurrently, the production of 15NO2⁻ and 15NO3⁻ was detected, indicating oxidation of 15NH4+ to 15NO2⁻ and 15NO3⁻ by ammonia-oxidizing microorganisms and nitrifiers in the biofilm. A decline in 15NO2⁻ and 15NO3⁻ concentrations was also observed, demonstrating the occurrence of denitrification during incubation. This finding was corroborated by the high abundance of denitrification genes detected (Fig. 3b).

Fig. 3: Dynamics of gene copy numbers and nitrogen transformation during stable isotope labeling cultivation.
Fig. 3: Dynamics of gene copy numbers and nitrogen transformation during stable isotope labeling cultivation.
Full size image

Changes in copy numbers of key functional genes driving ammonia oxidation, nitrification, and denitrification were detected during isotopic labeling cultivation of sample S3 (a, b). Copy number changes of the amoA gene in fragments after labeled culture on day 3 (c). The transformation of nitrogen in the culture medium over time (d).

Nitrogen metabolism processes driven by genes detected in heavy 15N-DNA fractions

Nitrogen metabolic genes in the metagenome were annotated using the NCycDB database, revealing genes associated with 5 key nitrogen cycling pathways (Fig. 4a and Table S5). In the nitrogen fixation pathway, sequences annotated as nifH and nifD genes were most abundant, indicating their role as key drivers of nitrogen fixation into biofilms on the stone material of the Former Residence of Chen Cihong. For ammonia oxidation and nitrification pathways, critical genes amoABC and nxrAB were annotated. For the reduction of NO3⁻ to NO2⁻, a critical step in denitrification, 15 genes driving dissimilatory nitrate reduction and 6 genes for assimilatory nitrate reduction were annotated in the 15N-enriched fractions. Genes encoding assimilatory nitrate reductase (NR, nasAB, narB) exhibited twice the abundance of those encoding dissimilatory nitrate reductase (narGHIJVWYZ, napA), indicating substantial activation of the assimilatory nitrate reduction pathway on day 3 of labeling and incubation.

Fig. 4: Metagenomic analysis of nitrogen-cycling genes in labeled microbial biofilms.
Fig. 4: Metagenomic analysis of nitrogen-cycling genes in labeled microbial biofilms.
Full size image

Heatmap analysis of the relative abundance of genes across distinct heavy 15N-DNA fractions (a). Nitrogen cycling network in biofilm colonizing stone material at Former Residence of Chen Cihong. The gray dashed line represents a possible anaerobic ammonia oxidation process, derived from 16S rDNA sequencing data (b).

Though both the key genes nirA (assimilation type) and nrfA (dissimilation type) involved in the nitrite reduction pathway were annotated, the substantial decrease in 15NH4+, 15NO2⁻ and 15NO3⁻ may suggest activity of the denitrification reaction. Among genes encoding two distinct types of nitrite reductase (nirK and nirS), sequences annotated as nirK gene were sixfold more abundant than those of nirS gene. This suggests that copper-containing nitrite reductase (Cu-NiR) encoded by nirK gene is the primary driver of nitrite reduction in the stone biofilm. Concurrently, norB, norC and nosZ genes in the denitrification pathway mediate the stepwise reduction of NO to N2. Collectively, these nitrogen-related genes collectively drive key nitrogen metabolic pathways in biofilm on the stone material of the Former Residence of Chen Cihong. Within this biofilm, ammonia oxidation, nitrification, and DNRA form an internal nitrogen loop (Fig. 4b), continuously generating NO2⁻ and NO3⁻ that allow conservation of nitrogen and also contribute to biodeterioration of stone material (Fig. 6).

Potential interconnected nitrogen and sulfur cycles in biofilm

Metagenomic analysis yielded 49 medium-to-high-quality MAGs (71.29–100% completeness; 0–9.99% contamination) spanning seven phyla (Table S6). Functional annotation of biogeochemical cycling genes using NCycDB and SCycDB databases identified 37 nitrogen cycling genes (Fig. S6) and 140 sulfur cycling genes (Fig. S7). Of these, a subset of 20 MAGs was of particular significance, as they co-harbored genes for both denitrification and sulfur oxidation (Fig. 5), highlighting their potential role in coupling nitrogen and sulfur cycling. These MAGs were phylogenetically diverse, belonging to Alphaproteobacteria, Gammaproteobacteria, and Bacteroidia. Notably, analysis of the denitrification pathways revealed a pattern of interdependence within the biofilm community; no single MAG encoded the complete pathway from NO3⁻ reduction to N2. Specifically, the capability for NO2⁻ reduction to NO (via nirK, nirS genes) and N2O reduction to N2 (via nosZ gene) was distributed among different MAGs. Only 2 MAGs (3-10_bin.25 and 3-12_bin.19) possessed both nirK or nirS and nosZ, suggesting a capacity for more complete denitrification. Additionally, 3 MAGs were annotated with narG gene (nitrate reductase) and nrfC gene (DNRA-associated), indicating potential capability for NO3⁻ reduction to NH4+, DNRA. This genomic evidence suggests that the complete denitrification process likely occurs through cross-feeding and synergistic interactions among different taxa within the biofilm, rather than being accomplished by a single organism. Furthermore, the metabolic potential for sulfur oxidation was widespread and versatile among these 20 MAGs. All contained core genes of the SOX system (soxB, soxXY genes), enabling thiosulfate (S2O32⁻) oxidation. Moreover, 7 MAGs possessed additional genes for oxidizing various inorganic sulfur compounds, including sulfide (sqr gene), elemental sulfur (dsrA gene), and SO32⁻/APS (aprAB, sat genes) (Table S7). This genetic repertoire indicates a capacity to utilize a broad range of sulfur compounds as electron donors.

Fig. 5: Metabolic characteristics of 20 MAGs on denitrification and sulfur oxidation in biofilm colonizing stone material at the Former Residence of Chen Cihong.
Fig. 5: Metabolic characteristics of 20 MAGs on denitrification and sulfur oxidation in biofilm colonizing stone material at the Former Residence of Chen Cihong.
Full size image

Blue arrows indicate denitrification reactions and purple arrows indicate sulfur oxidation reactions.

The coupling between incomplete denitrification and sulfur oxidation is primarily manifested in processes such as the reduction of NO2⁻/NO3⁻ to NO, the reduction of N2O to N2, coupled with the oxidation of H2S/S2O32⁻/APS to SO42⁻, as exemplified by Pseudomonas (3-9_bin.8) and Flavobacterium (3-11_bin.6, 3-11_bin.40, and 3-12_bin.25). Collectively, the co-occurrence of denitrification and multi-step sulfur oxidation genes in these diverse MAGs provides genomic evidence supporting a synergistic mechanism. In this proposed model, sulfur compounds serve as electron donors for energy generation, while nitrogen oxides (e.g., NO2⁻, NO3⁻, N2O) from nitrification act as terminal electron acceptors for denitrification. This coupling not only fuels microbial metabolism but also drives the acid production (from sulfur oxidation) and salt stress (from nitrate transformation), which synergistically accelerate the deterioration of the stone substrate (Fig. 6).

Fig. 6: Schematic diagram showing biofilm-driven acid attack mechanisms associated with nitrogen and sulfur cycling on stone cultural heritage at the Former Residence of Chen Cihong.
Fig. 6: Schematic diagram showing biofilm-driven acid attack mechanisms associated with nitrogen and sulfur cycling on stone cultural heritage at the Former Residence of Chen Cihong.
Full size image

In biofilms, diazotrophic bacteria fix atmospheric N2 into NH4+, which serves as a nitrogen source for ammonia-oxidizing microorganisms. Subsequently, NO3⁻ generated from nitrification exhibits acid attack potential on stone substrates. The fate of NO3⁻ primarily involves two transformation pathways: 1) Internal nitrogen loop via DNRA, coupled with ammonia oxidation and nitrification; 2) Nitrogen loss through denitrification. Furthermore, NO3⁻ (as an electron acceptor) links denitrification with sulfur oxidation, generating highly acidic corrosive SO42⁻. During wet phases, NO3⁻ and SO42⁻ migrate with water into stone fissures, forming inorganic salts (e.g., CaSO4, MgSO4) with cations such as Ca2+ and Mg2+, While during dry phases, crystallization pressure from salt precipitation accelerates stone crack propagation. This internal nitrogen cycling, coupled with the nitrification-denitrification-sulfur oxidation mechanism, persistently drives acid attack and salt damage on stone substrates at the Former Residence of Chen Cihong.

Discussion

The deterioration inflicted by processes within the nitrogen cycle on stone cultural heritage is both persistent and elusive. In this study at the Former Residence of Chen Cihong, an analysis of microbial communities and nitrogen-cycling genes across different building materials revealed significant populations of ammonia-oxidizing microorganisms on stone surfaces. This finding aligns with functional gene-based studies predicting the involvement of element-cycling bacteria in the corrosion of historical stone structures, which identified nitrogen cycle processes encompassing ammonia and nitrite oxidation50. Notably, in this study, AOA exhibited higher abundance than AOB, particularly in the high-floor sample S3. Due to their broader ammonia affinity compared to AOB, AOA adapt to low substrate concentration environments and dominate51,52,53. Consequently, even on nutrient-poor stone heritage surfaces, AOA abundance typically exceeds that of AOB, as observed in Bayon Temples of Cambodia18,54 and many other stone cultural heritage21. Nitrifying bacteria exhibited high abundance in both samples S1 and S3. Furthermore, stable isotope tracing of the S3 biofilm confirmed the production of nitrification products (15NO2⁻ and 15NO3⁻), demonstrating active nitrification processes in biofilm on the stone material of the Former Residence of Chen Cihong. It should be noted that while DNA-SIP is a powerful tool for linking microbial identity to function, potential limitations such as the possibility of indirect label incorporation (cross-feeding) through trophic interactions must be acknowledged. The interpretations in our study are based on an abundance of functional gene and parallel metabolic evidence to mitigate this potential limitation. Ammonia oxidation is the rate-limiting step in nitrogen cycling55 and is tightly coupled with nitrite oxidation. Therefore, the lower acidity observed in sample S3, compared to S1, may be explained by differences in the abundance and activity of ammonia-oxidizing microorganisms, which ultimately influence nitrification efficiency. NO2⁻ and NO3⁻ produced during nitrification pose dual threats to stone cultural heritage through acid dissolution and salt crystallization damage. They dissolve carbonate minerals within stone materials directly, leading to surface powdering and spalling9, dissolved cations (e.g., Ca2+, Mg2+) associate with NO3⁻ to form soluble nitrate salts. Though this study did not employ direct mineralogical analysis to identify the specific types of nitrate salts formed, the presence of active nitrification processes in the stone, coupled with the detection of NO3⁻ via ion chromatography, strongly indicates the in situ formation of nitrates, which migrate into and out of the stone matrix under water cycling conditions. Upon humidity reduction or drying, salt crystallization exerts considerable internal pressure, causing physical cracking damage to the stone56.

The impact of denitrification on stone cultural heritage may exhibit both potential protective and deteriorative effects. Denitrification reduces NO3⁻ to N257, consuming NO3⁻ and thereby eliminating the substrate for subsequent DNRA and the total salt load available for crystallization damage. This process also disrupts the strongly acidic environment dominated by nitrification, mitigating persistent acid attack and yielding a positive protective effect on stone heritage. Nitrification typically occurs on oxygen available stone surfaces, whereas denitrification operates in micro zones where oxygen is depleted57,58,59,60, such as deeper pores, internal fissures, or anoxic layers within biofilm. This spatial segregation implies limited interaction between the two processes. Though nitrification-generated NO3⁻ on stone surfaces can migrate inward via moisture diffusion movement3, incomplete denitrification frequently occurs due to truncated metabolic pathways and is constrained by multiple factors (e.g., temperature and pH), resulting in accumulation of intermediates such as NO2⁻ or N2O61,62. Therefore, the denitrification process cannot be simply regarded as a protective mechanism for stone cultural heritage. Conversely, denitrification consumes H+, thereby elevating local pH within microenvironmental niches. Silicates form the fundamental framework of many rocks, such as granite. Mineralogical studies indicate that prolonged exposure to alkaline conditions can disrupt their silicon-oxygen network, leading to mineral dissolution and the generation of soluble silicates. This process results in the formation of micropores and pits on the stone surface, thereby compromising its structural integrity63,64. Consequently, the dual impact of denitrification on stone cultural heritage leans toward being predominantly deteriorative.

DNRA is an essential nitrogen conversion pathway that serves as a critical node linking ammonia oxidation and nitrification to establish an internal nitrogen loop14. Our study identifies active genes of DNRA processes on stone substrates of the Former Residence of Chen Cihong. This pathway utilizes NO3⁻, which is partially derived from ammonia oxidation or nitrification, as a substrate. The regenerated NH4⁺ then provides a renewed substrate for ammonia-oxidizing microorganisms, thereby driving the continuous regeneration of NO3⁻ and acidity through the ammonia oxidation reaction. Long-term NO3⁻ accumulation has been documented on stone cultural heritage in Cambodia, China, and European countries21,65,66. Specifically, high-concentration NO3⁻ accumulation detected in sandstone temples of Angkor monuments is directly attributed to the metabolic activity of AOA, elevating NO3⁻ content significantly above ambient background levels. DNRA microorganisms reduce the concentration of NO3⁻ and supply NH4+, which ammonia-oxidizing microorganisms further oxidize to NO3⁻. The resulting NO3⁻ is associated with Ca2+ in stone matrices to form calcium nitrate, Ca(NO3)2.

This persistence of the internal nitrogen loop poses long-term structural deterioration risks to stone heritage. Moreover, its impact on stone degradation is intrinsically linked to denitrification processes. As demonstrated by Fan et al.67, denitrification facilitates NO3⁻ removal while regulating DNRA, thereby modulating internal nitrogen loop dynamics. In contrast to sandstone Angkor monuments in Cambodia, active denitrification was observed on stone surfaces at the Former Residence of Chen Cihong. Crucially, consumption of NO3⁻ through denitrification not only disrupts acidic environments generated by ammonia oxidation and nitrification but also constrains DNRA activity. Additionally, low-abundance anaerobic ammonia-oxidizing bacteria detected in S3 partially consume NO2⁻ and NH4+, generating N2 via anammox reactions, potentially further weakening the internal nitrogen loop. Collectively, although the internal nitrogen loop persists at the Former Residence of Chen Cihong, its reactions are regulated by denitrification and anammox processes.

Microbial nitrogen cycling is the core pathway for nitrate (NO3⁻) accumulation on stone cultural heritage. In sulfur-rich environments, these microbial processes exhibit even greater metabolic diversity68. More importantly, the coupling of nitrogen cycling with sulfur oxidation can generate synergistic effects that significantly accelerate stone deterioration. In situ Analyses of building materials at the Former Residence of Chen Cihong, particularly stone materials, revealed the presence of both NO3⁻ and SO42⁻. Stable isotope labeling combined with metagenomic analysis further identified multiple MAGs harboring genes for denitrification and sulfur oxidation, indicating microorganisms with potential for coupled denitrification-sulfur oxidation processes on stone surfaces. These reactions utilize reduced sulfur as electron donors69,70 and are particularly active in low C/N environments71,72. Although denitrification removes NO3⁻, NO2⁻ accumulation has been frequently observed in sulfur-based denitrification due to an imbalance of denitrification and denitrification rate73,74,75,76, potentially explaining the high abundance of 15NO2⁻ observed during early cultivation. The significant accumulation of SO42⁻ detected by ion chromatography (IC) on the stone cultural heritage, combined with the annotation of sulfur oxidation genes (e.g., those encoding the SOX system, sat, aprA, fccB, sqr, and dsrAB genes) from stable isotope-labeled metagenomic datasets, provides strong evidence for microbially driven sulfur oxidation in situ. Although direct mineralogical identification of sulfate salts (e.g., gypsum) was not the focus of this study, the co-occurrence of SO42⁻ levels and functional genetic potential strongly suggests a high risk of salt crystallization damage21. Furthermore, studies have demonstrated that the generation of elemental sulfur as an intermediate byproduct may form a positive feedback loop, promote the growth and activity of sulfur-oxidizing bacteria and thereby intensify acid production and chemical weathering77. In summary, integrated evidence from metagenomic and stable isotope probing suggests that the coupling of nitrogen and sulfur cycles on stone cultural heritage is likely predominantly mediated by a nitrification-denitrification-sulfur oxidation mechanism (Fig. 6). Although the direct decay rate of the stone was not measured, this synergistic process may serve as a key driver of chemical weathering and physical deterioration of the stone substrate by enhancing acid production efficiency. Therefore, combining our experimental evidence with established biogeochemical theory supports the view that microbe‑mediated biogeochemical metabolic networks are important contributors to the deterioration of stone heritage.

The stone material of the Former Residence of Chen Cihong, along with most large-scale outdoor cultural heritage sites, is susceptible to acid attack driven by microbially mediated biogeochemical cycling. Therefore, protection strategies must be tailored to these specific deterioration mechanisms. For the sustainable conservation of such heritage, future strategies could target key metabolic pathways through: i) the cautious application of chemical inhibitors to suppress the activity of ammonia-oxidizing and DNRA microorganisms, or ii) the implementation of established biocleaning methods to remove accumulated NO3⁻, thereby disrupting the nitrogen cycle. However, the practical application of these strategies faces several limitations. The efficacy and potential side effects of chemical inhibitors, such as their environmental persistence, impact on non-target microbial communities, and the need for periodic reapplication in changing field conditions, require thorough pilot testing and long-term monitoring prior to large-scale implementation. Similarly, biocleaning protocols must be optimized for diverse and complex real-world scenarios to ensure effectiveness and safety. More importantly, future research should integrate multi-omics technologies with in situ monitoring to resolve the spatiotemporal dynamics of nitrogen-sulfur coupled reactions. This integrated approach will be fundamental to developing reliable, microbial ecology-based “targeted interception” strategies that can effectively mitigate biodeterioration while ensuring the safety of the cultural heritage materials.