Ni²⁺-Assisted Hydrolysis May Affect the Human Proteome; Filaggrin Degradation Ex Vivo as an Example of Possible Consequences

Amino Acid Enrichment

Experimental amino acid occurrences on position X₁ of X₁-S/T-c/p-H-c-X₂ motif were corrected for the abundance of each amino acid in the whole proteome. Positive value indicates that a particular residue type is found on X-position more frequently than in the whole proteome.

Cumulative Distribution Function (CDF)

Cumulative distribution functions were obtained according to the standard approach (Montgomery and Runger, 2007) and visualized using Origin software (version 9.7, www.originlab.com).

Quantitative Analysis of Hydrolytic Motifs Frequency Within the Human Proteome

The number of X₁-S/T-c/p-H-c-X₂ motifs was determined for each individual protein. The distribution of these numbers was then analysed assuming Poisson distribution, which assumes no correlations between particular motifs. The analysis was performed using Origin software (version 9.7, www.originlab.com).

Single-Term GO Functional Enrichment Analysis

Gene Ontology functional enrichment analysis was performed using topGO R package with custom GO mapping files based on Gene Ontology human annotations files from April 2018. Only Gene Ontology terms from Biological Process were used for annotation. Initial datasets G-S/T-c/p-H-c-X₂ and X₁-S/T-c/p-H-c-X₂ included 4111 and 31099 proteins denoted with UniProt accession numbers. As GO annotations are provided for UniProtKB identifier, the first step involved translation UniProt accession numbers to Uniprot KB identifiers and filtering out isoforms. After the filtering step, the datasets G-S/T-c/p-H-c-X₂ and X₁-S/T-c/p-H-c-X₂ 2938 and 23120 protein identifiers respectively. To assess the statistical significance of GO terms enrichment in both datasets, Fisher Exact test was used with Benjamin-Hochberg adjustment for multiple testing corrections.

Analysis of Hydrolytic Motifs Within Filaggrins From Various Species

In order to create the list of proteins from diverse organisms we used the ScanProSite Tool which allows to scan a protein database against a motif. We used the X₁-S/T-c/p-H-c-X₂ motif for this purpose. We then normalized the number of cleavage sites dividing it by the total sequence length (# of motifs * 100/seq length). Dataset has been used to create a circular phylogenetic tree in NCBI CommonTree (https://www.ncbi.nlm.nih.gov/Taxonomy/CommonTree/wwwcmt.cgi) and visualized in iTOL https://itol.embl.de/(Letunic and Bork, 2021).

In Silico Analysis of proFLG Sequence and Selection of Peptides

In silico analysis of proFLG sequence was performed on the base available in NCBI database human proFLG amino acid sequence (Ref. {"type":"entrez-protein","attrs":{"text":"NP_002007.1","term_id":"60097902","term_text":"NP_002007.1"}}NP_002007.1) and information about FLG repeats (Sandilands et al., 2009). The FLG domains separation was proposed and compared using WebLogo 3 application (Schneider and Stephens, 1990; Crooks et al., 2004).

Peptide Synthesis

All peptides were synthesized in the solid phase according to the Fmoc protocol (Chan and White, 2000) using the Prelude automatic synthesizer (Protein Technologies). The syntheses were accomplished using N-α-9-Fluorenylmethyloxycarbonyl (F-moc) amino acids (Novabiochem) on a TentaGel^® S RAM resin (Rapp Polymere), using O-(Benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU, Sigma-Aldrich) as a coupling reagent, in the presence of N,N-diisopropylethylamine (DIEA, Sigma-Aldrich). The acetylation of the N-terminus was carried out in 10% acetic anhydride in DCM. The cleavage was done manually by the cleavage mixture composed of 95% trifluoroacetic acid (TFA, Sigma-Aldrich), 2.5% triisopropylsilane (TIS, Sigma-Aldrich) and 2.5% water. Peptides were isolated from cleavage mixtures by the addition of ice-cold diethyl ether and centrifugation. Following precipitation, peptides were dissolved in water and lyophilized. Peptides were purified by HPLC (Waters) using an analytical C18 column (ACE 250 × 4.6 mm) monitored at 220 and 280 nm. The eluting solvent A was 0.1% (v/v) TFA in water, and solvent B was 0.1% (v/v) TFA in 90% (v/v) HPLC grade acetonitrile (Rathburn Chemicals). The correctness of molecular masses and purities of the peptides was confirmed using Premier Q-Tof ESI-MS spectrometer (Waters). After this step, peptide solutions were frozen in liquid nitrogen and lyophilized.

Ni²⁺-Assisted FLG Peptides Hydrolysis

The hydrolysis experiments were performed in a 20 mM HEPES buffer (Sigma-Aldrich), using 0.5 mM peptide and 2 mM Ni(NO₃)₂ (Sigma-Aldrich). The samples were incubated in pH 8.2, at 50°C and pH 7.4, at 37°C. The aliquots were periodically collected from the samples and acidified by addition 2% (v/v) TFA. Control samples, containing peptide and buffer, but without Ni²⁺, were gathered at the same time points. For analysis, reaction mixtures were diluted by water 4 to 1 and injected into the HPLC system (Waters), equipped with an analytical C18 column. The eluting solvent A was 0.1% (v/v) TFA in water, and solvent B was 0.1% (v/v) TFA in 90% (v/v) acetonitrile. The chromatograms were obtained at 220 and 280 nm. After separation, the products of hydrolysis were identified using electrospray ionization mass spectrometry (ESI-MS). The relative amounts of these fractions in each chromatogram were calculated by peak integration using data analysis software Origin 8.1 or Origin Pro 2017 (OriginLab Corporation).

Kinetic Analysis

To calculate the rate constants of the acyl shift step (k ₁) and the ester hydrolysis step (k ₂) of the hydrolysis reaction the set of three equations (Kinet A, Kinet B, and Kinet C) was used, similarly to previous studies (Kopera et al., 2010; Ariani et al., 2013; Protas et al., 2013).

In these equations y is a molar fraction of a given species, x is the time axis, and A₀ denotes the initial concentration of the substrate.

UV−Visible and Circular Dichroism Spectroscopies

The UV-visible spectra were recorded in the range of 850–330 nm, on a LAMBDA 950 UV/vis/NIR spectrophotometer (PerkinElmer). The path length was 1 cm. Complexometric titrations were performed for the samples containing 0.95 mM peptide and 0.9 mM Ni(NO₃)₂ dissolved in H₂O. The pH of the solution was adjusted manually in the range of 3–11.5 by titrating with small amounts of concentrated NaOH. Circular dichroism (CD) spectra of Ni²⁺ complexes with peptides were recorded in the range of 270–800 nm, on a Jasco J-815 spectropolarimeter, using the same samples as for UV-vis experiments. The pK _a values for the complex formation were obtained by fitting the absorption value at the band maximum to the Hill equation (Acerenza and Mizraji, 1997).

Molecular Modeling of Ni²⁺ Complexes

All molecular mechanics simulations were performed using YASARA2 force-field (Krieger et al., 2004) extended for the Ni²⁺ coordination by adding the ab initio derived topology and charge distributions. The N-Ni distances were constrained as a pseudo-bond of the appropriate length, and the geometry was forced according to the expected square planar coordination of Ni²⁺ by additional constrains for N-Ni-N angles (90°) and N-Ni-N-N pseudo-dihedrals (180°). Additional pseudo-dihedral constrains were introduced to mimic sp² hybridisation of the Nitrogen (C-N-Cα-N = 180°). Model peptides were built in extended conformation and the structure of their complexes with Ni was initially optimised using implemented in Yasara algorithm that combines simulated annealing and energy minimisation. The further ten cycles of high temperature molecular dynamics (250 ps at 1,000 K) followed by stepwise cooling and final energy minimization were done to assess conformational flexibility of a given Ni-peptide complex. Molecular graphics were created with YASARA (www.yasara.org) and POVRay (www.povray.org).

FLG Recombinant Protein: Plasmid Construction

Plasmids were constructed using a sequence- and ligation-independent cloning (SLIC) method (Li and Elledge, 2007). Nucleotide sequence encoding 10th FLG repeat domain were encloned in pET28 vector using BamHI and XhoI restrictions sites. The FLG 10th construct contained a C-terminal His6-tag. As a control, a construct for maize protein kinase CK2α was obtained in similar conditions. The CK2α protein has a molecular weight similar to FLG and contains no X₁-S/T-c/p-H-c-X₂ motifs. The maize protein kinase CK2α construct contained a C-terminal His6-tag. The constructs were verified by sequencing.

Protein Production and Purification

The constructs were transformed into E. coli BL21-CodonPlus-RIL and propagated overnight in LB liquid media containing kanamycin and chloramphenicol at 37°C. The bacterial cultures were diluted 1:100 in LB liquid media supplemented by antibiotics and incubated at 37°C until the culture has reached the mid-log phase of growth. Protein expression was induced by IPTG (1 mM) for 2 h at 37°C. The cells were harvested by centrifugation (10 min, 5,000 × g, 4°C). The pellets were mixed with lysis buffer (10 mM Tris pH 8, 150 mM NaCl, 10 mM imidazole) supplemented with protease inhibitors cocktail and lysed by sonication. The cell lysate was clarified by centrifugation (60 min, 24,000 × g, 4°C) and used for affinity purification on a HisPur™ Cobalt Resin (Thermo Fisher). Pure protein was eluted by an elution buffer (10 mM Tris pH 8, 150 mM NaCl, 300 mM imidazole). Samples were dialysed (10 mM Tris pH 8.5, 150 mM NaCl, MWCO: 12–14000 Da) and analyzed on SDS-PAGE. Bands of interest were cut out and identified by MALDI-TOF MS after trypsin digestion.

Ni²⁺-Assisted FLG Monomer Hydrolysis

FLG protein domains (30 µM) were incubated in 10 mM TRIS/150 mM NaCl buffer with or without nickel ions [1 mM Ni(NO₃)₂] under optimal (pH 8.2, 50°C) and physiological (pH 7.4, 37°C) conditions. The reactions were stopped by freezing the collected samples in liquid nitrogen. Samples from different time points were separated using the Tricine-SDS page technique and Bio-Rad system. The experiment was repeated for CK2α control protein. Gels after electrophoresis were scanned (E-gel imager Camera, Life Technologies) and the scans used for densitometric analyses (ImageJ program).

Cell Proliferation Assay

In order to find out the toxic concentration of Ni(NO₃)₂ for keratinocytes, the cell proliferation assay (MTT) was performed. Cells were cultured in a 96-well plate and after 24 h of exposure to a gradient of Ni(NO₃)₂ concentrations (10⁻² to 10⁻⁷ M final concentration) the test was performed according to manufacturer’s protocol (CellTiter 96^® Non-Radioactive Cell Proliferation Assay, Promega). The assay determined the half maximal inhibitory concentration value IC₅₀ as approximately 1 mM Ni(NO₃)₂.

Normal Human Epidermal Keratinocyte Culture

Normal human epidermal keratinocytes (NHEKs) (purchased from Lonza) were cultured in monolayers in a dedicated medium (Lonza, KBM-2) at the Ca²⁺ level of 0.06 mM. To stimulate differentiation and FLG expression, a calcium switch was conducted over a period of 24 h by replacing the culture media with fresh media adjusted to a 1.5 mM final calcium concentration. A Ni(NO₃)₂ (Sigma) solution was added to achieve various final concentrations (10 μM, 100 μM and 1 mM). Doses were chosen based on MTT test results (Supplementary Figure S1). After 24 h of incubation, the cells were fixed, permeabilized and immunostained with anti-FLG antibodies (Anti-FLG goat G20 (Santa Cruz), and secondary anti-goat Alexa-488 and anti-rabbit Alexa-568 (Life Technologies) antibodies were used. Staining was carried out in PBS and nuclei were visualized by Hoechst (NucBlue, Life Technologies). The slides were coversliped with Mowiol 4-88 (Sigma). Data acquisition was carried out on the Zeiss 780 inverted confocal microscope. Images from three separate experiments were analysed; KHG diameter and integrated intensity from the signal were measured using Fiji: ImageJ program (Abramoff, 2007). For the statistical analysis the Mann–Whitney U test was used.

Exposure of Epidermal Sheets to Nickel

Skin samples were obtained from healthy donors undergoing surgery under ethical approval from the UK National Research Ethics Service (14.NW.1153). Epidermal sheets were separated from dermal tissues by overnight incubation in dispase (5 U/ml; Sigma Aldrich) and cultured up to 48 h in KGM-2 keratinocyte medium (Lonza) adjusted with CaCl₂ to a 1.5 mM final calcium concentration. The Ni(NO₃)₂ solution was added at the 1 mM final concentration. Experiments were repeated on skin explants from 10 donors. For fluorescent antibody staining epidermal sheets were fixed with 4% formaldehyde (Sigma), followed by 0.1% Triton X-100 (Sigma Aldrich) and incubated in a blocking buffer (5% FCS, 2% BSA in PBS) for 1 h. Anti-FLG goat G20 (Santa Cruz), and the secondary anti-goat Alexa488 (Life Technologies) antibody staining was carried out in PBS for 1 h. The nuclei were visualized by Hoechst (NucBlue, Life Technologies). The sheets were mounted on microscope cover-slides with Mowiol 4-88 (Sigma) for imaging. Data acquisition was carried out on the Zeiss 780 inverted confocal microscope by recording z-stacks of 2D images (at 0.38 µM intervals) and images taken using inverted confocal microscope (Zeiss 780) by recording 2D images in a 3D z-stack.

Western Blot Analysis

Isolated epidermal sheets were washed in PBS and incubated in an 8M urea buffer (ReadyPrep™ Sequential Extraction Kit, Reagent 2 with reducing reagent; Bio-Rad) and sonicated in a water bath for 30 min. Lysates were spun at 4°C (13,000 rpm, 15 min). Proteins from supernatants were fractionated 7% Tris-Acetate NuPage gels (Life Technologies). Proteins were transferred onto PVDF membranes (iBlot Dry Blot system stacks and iBlot transfer device; LifeTechnologies). Membranes were blocked in a 5% solution of non-fat milk powder in PBS and incubated overnight with desired primary antibodies (anti-FLG goat G-20 (Santa Cruz). Li-Cor infrared secondary antibodies and Li-Cor scanning system (Li-Cor Biosciences) were used for detection.

RNA Isolation and the Assessment of RNA Integrity

Isolation of total RNA from N-HEK cells was performed using the PureLink^® RNA Mini Kit (Ambion) according to the manufacturer’s protocol. The RNA concentration and purity was estimated using NanoDrop 2000 (Thermo). The assessment of RNA quality was carried out on the Agilent 2100 Bioanalyzer System and Eukaryote Total RNA Nano Assay kit (Agilent) was used, according to the manufacturer’s protocol.

Reverse Transcription

Isolated RNA was used as a template for the reverse transcription. The reaction was performed in a S1000™ Thermal Cycler (Bio-Rad) in 20 µl using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystem) according to the manufacturer’s protocol. The ingredients contained: 1,000 ng RNA per sample, reaction buffer, random primers, mix of dNTPs, RNase inhibitor (1.0 U/μl) and MultiScribe™ reverse transcriptase (2.5 U/μl). Conditions of RT reaction are presented in the table below.

Quantitative Real-Time PCR

Gene expression assay was counducted using the StepOne™ Real-Time PCR System and the TaqMan^® Gene Expression Assay (Applied Biosystems). Briefly the reaction (holding stage I: 2 min, 50°C; holding stage II: 5 min, 95°C; Cycling stage (40x): 15 s, 95°C and 1 min, 60°C) was performed in 10 µl total volume with 12.5 ng of cDNA (or water as a negative control) addition. TaqMan^® Universal Master Mix II with the AmpErase^® UNG (uracil-N-glycosylase) (Applied Biosystems) was used. The primers (forward: 5′–GGA​AAA​GGA​ATT​TCG​GCA​AAT–3′, reverse: 5′–TCC​ATG​AAG​ACA​TCA​ACC​ATA​TCT​G–3′) and the TaqMan^® MGB probe (5′–FAM CTGAAGAATCCAGATGAC-NFQ-MGB–3′, Applied Biosystems) set for FLG gene was designed using Primer Express software (Applied Biosystems) and the specificity was checked using PrimerBLAST tool (NCBI). The specificity of the primers and probe set was confirmed by adequate negative controls. Results were calculated based on a ΔC_T method and TBP1 (TATA-box binding protein 1) gene (commercially available primer and probe set, accession number: Hs00427620, Applied Biosystems) was used as a reference. For statistical analyses the t-test was used.

Langerhans Cell-like Cells

Peripheral blood mononuclear cells (PBMCs) were isolated from blood collected from healthy adult donors under local ethics approval (09/H0606/71). Samples were diluted and centrifuged in a density gradient using a Lymphoprep™ reagent (STEMCELL Technologies Inc.). The CD14⁺ cells were separated with a MACS MicroBead (Miltenyi Biotec) magnetic separation system according to the manufacturer’s protocol. Subsequently, the monocytes were cultured in a 1 × 10⁶/ml density in the RPMI medium supplemented with 10% fetal calf serum, penicillin/streptomycin mix and 2 mM L-glutamine, with addition of cytokines: 250 ng/ml GM-CSF, 100 ng/ml IL-4, 10 ng/ml TGF-β1, all obtained from Pepro-Tech. After 5 days of culture, the cells were exposed to a Ni(NO₃)₂ solution (1 mM), peptides (50 µM) or Ni²⁺-peptide complexes with the same concentrations of peptides and of the nickel salt. After 48 h the cells were harvested for the flow cytometry analysis.

Flow Cytometry

The cells were harvested by decantation into a conical tube and centrifuged (10 min, 1,400 rpm, 4°C). Supernatants were collected and frozen until further analysis. Next, the cells were washed in ice cold 10% FCS in PBS and stained. All staining was carried out on ice and protected from light. Conjugated primary antibodies: anti-human CD86 (APC) and HLA DR (PE), CD80 (FITC) were added in 0.1–10 μg/ml concentration range and incubated for 1 h in the dark at 4°C. The cells were washed three times in PBS and centrifuged (5 min, 1,400 rpm, 4°C), and resuspended in 1 ml of ice cold PBS, containing 10% FCS and 1% sodium azide. The cells were fixed in 1% paraformaldehyde solution and kept in the dark on ice until the analysis. The cytometric analysis was performed on CyAn™ ADP (Beckman Coulter). First, using unstained cells and compensation beads (Anti-Mouse Ig, κ/Negative Control Compensation Particles Set, BD), the compensation procedure was performed. The FCS Express 7 Flow Cytometry Software—RUO, DeNovo Software were used for the final analysis. The results were analysed with the t-test.

Ni-Hydrolysis Occurs in Filaggrin Model Peptides and the Recombinant Filaggrin Monomer Domain

The barrier proteins abundant in the Ni-hydrolytic motifs are potential targets for Ni²⁺ ions and their concomitant hydrolysis might compromise their function. We chose FLG for the in vitro and ex vivo experimental verification of this hypothesis due to its largest number of G-motifs (67 in proFLG) and its importance in protection from allergic sensitization, including to nickel (Novak et al., 2008). We also searched our database for other proteins related to keratinocyte differentiation. Interestingly, lorricin, involucrin, trichohyalin and elafin do not contain any hydrolytic motifs, while the previously mentioned filaggrin, filaggrin-2 and hornerin are enriched with such motifs (Supplementary Table S14).

The FLG monomer domains were identified on the basis of the amino acid sequence of human proFLG, (NCBI database Ref. {"type":"entrez-protein","attrs":{"text":"NP_002007.1","term_id":"60097902","term_text":"NP_002007.1"}}NP_002007.1) and a prior study on the proFLG component domains (Sandilands et al., 2009). The resulting domains were compared using the WebLogo 3 application. A typical FLG monomer domain contains 17 potential Ni-hydrolytic sites including 7G-motifs with a high degree of conservation, as shown in Figure 2A. Next, we chose eight oligopeptides containing Ni-hydrolytic motifs best representing the cleavage sites, taking into account the variability of third and fifth positions in the X₁-S/T-c/p-H-c-X₂ sequence (Supplementary Table S15). The abbreviation numbers of filaggrin peptides (FP) denote the order of their occurrence in the monomer (Figure 2A). Molecular modelling of ten structures of Ni²⁺ complexes with these peptides characterised by the lowest calculated energy levels repeatedly showed square-planar structures with Ni²⁺ chelate ring conformations very similar to each other (Figure 2B, Supplementary Figure S2). Next, comparative CD and UV-vis spectroscopic pH titrations were performed in a broad pH range, yielding the pH dependence of complexation (Figure 2C, Supplementary Figure S3). The calculated pK values showed the absence of stable square-planar complexes below pH 7 for all peptides. Since the Ni²⁺ binding at the individual hydrolytic motifs depends solely on the local sequence, and most of these sites are sufficiently separated from each other, this property of model peptides could be extrapolated over the entire protein (Kopera et al., 2010; Krezel et al., 2010).

FIGURE 2

FLG structure and susceptibility to Ni-hydrolysis. (A) The FLG gene, located on chromosome 1q21, consists of three exons and two introns. ProFLG protein contains several nearly identical FLG monomer units (green) flanked by partial monomer repeats and ...

Subsequently, we studied the kinetics of FPs hydrolysis as in our previous work (Kopera et al., 2010), using both harsh (50°C, pH 8.2) and physiological (37°C, pH 7.4) conditions. The peptides were hydrolysed in all cases (Figure 2C, Supplementary Figure S4). The kinetic parameters were calculated according to the model of the two sequential first order processes of the intermediate ester formation and decay into final products, as stipulated by the reaction mechanism (Krezel et al., 2010; Podobas et al., 2014). The values of k₁ and k₂ rate constants, describing these reaction steps, are presented in Supplementary Table S16. The reaction rates varied depending on the peptide sequence, and the hydrolysis was much faster in harsh conditions, as expected (Kopera et al., 2010).

Next, we confirmed the occurrence of Ni-hydrolysis for the FLG monomer domain, using the recombinant 10th FLG monomer domain (FLG-10, full sequence in Supplementary Table S17). The nickel concentration differed from that used in the peptide model experiments (2 and 1 mM respectively). Nickel hydrolysis has been the subject of extensive investigations in our research group. The conditions chosen for model oligopeptide studies corresponded to previously described experiments on similar peptides (Protas et al., 2013; Podobas et al., 2014; Wezynfeld et al., 2014). The conditions used for FLG domain had lower total Ni²⁺, but higher Ni²⁺/peptide ratio and were aimed at more accurate mirroring of the skin conditions. We would like to note that the Ni²⁺/peptide ratio is more relevant for the reaction rate than the absolute Ni²⁺ concentration, but the rate is ultimately controlled by the cleavage site saturation (Kopera et al., 2010). Recombinant maize protein kinase CK2α which has no Ni-hydrolytic sites (full sequence in Supplementary Table S18) served as a negative control (Figure 2D). Therefore, FLG cleavage resulted specifically from the Ni²⁺ presence rather than a residual protease activity. In order to compare the kinetics of the hydrolysis of FLG-10 vs. the FPs, the rate constants for the latter were recalculated by fitting the first order rate law to the final reaction product formation, as described previously (Krezel et al., 2010) (Supplementary Figure S5 and Supplementary Table S16). This was done since only the final reaction products could be quantified in protein gels, while the separate k₁ and k₂ values could be determined for the peptides for an excellent separation of the respective reaction products by HPLC approach (Podobas et al., 2014). The FLG-10 hydrolysis products showed a reproducible pattern of bands, i.e., initially, the two dominant masses (around 25 and 12 kDa) appeared, followed by subsequent hydrolysis of the 25 kDa fragment. The final hydrolysis products had masses within the range of 9–12 kDa, correlating with the cleavage primarily within FP-05 followed by FP-09 and FP-10. As presented in Figure 2D, the t_½ for the final product formation at harsh conditions (pH 8.2, 50°C) was ca. 20 min vs. 3.6 h at physiological conditions (pH 7.4, 37°C), both reactions proceeded according to the pseudo-first order rate law. The similar time evolution of gel band patterns at these two conditions indicated that the relative reaction rates at individual cleavage sites were maintained in FLG-10. The comparison of fragment sizes at the shortest incubation times with the pattern predicted from the sequence analysis and reactions of peptides confirmed FP-05 and FP-13 as initial reaction sites, followed rapidly by other sites; the entire protein was cleaved into small fragments within hours. Under harsh conditions the rate constant for the FLG domain decay is roughly equal to the sum of rates at the individual hydrolysis sites (Supplementary Figure S6A), while for the physiological conditions the FLG decay is several fold faster than one might expect from the model peptide data Supplementary Figure S6B (note that according to the reaction mechanism the rate constants for cleavages at different FLG sites add up to the overall rate of the domain decay). Altogether, we noticed that at physiological conditions the domain decayed ca. 10 times slower in comparison to the harsh conditions. The multiplicative effect of lowering the reaction temperature and pH on the reaction rate can be estimated as ca. 60–70, stemming from the temperature factor, ca. 2–2.5 and the pH factor, ca. 20–50 (Kopera et al., 2010); here we estimated ca. 40–50 for the most active peptides (Supplementary Figure S6C).

Ni-Hydrolysis of Filaggrin Occurs in Human Keratinocytes In Vitro and in Human Epidermis Ex Vivo

Having determined that the recombinant FLG monomer domain and its model peptides are cleaved by Ni-hydrolysis, we went on to investigate the biological meaning of this phenomenon at both the cellular and tissue levels. Since FLG is expressed predominantly in keratinocytes which are well differentiated, for the cellular study we used normal human epidermal keratinocytes, NHEKs, cultured in the differentiation-promoting medium, i.e. previously well-established calcium-switch model (Gutowska-Owsiak et al., 2018; Gutowska-Owsiak et al., 2020) (Figure 3A). These experiments determined that both the number and sizes of KHGs in NHEKs were reduced upon the Ni²⁺ treatment. Interestingly, some positive staining with anti-FLG antibodies could be observed as KHG-unrestricted cytoplasmic or filamentous signal, suggesting the release of the antibody-reactive FLG-derived peptides into the cytoplasm and potentially binding of those to the intermediate keratin filaments in accordance to the native function of FLG monomers.

FIGURE 3

FLG hydrolysis in human keratinocytes in vitro and human epidermis ex vivo. (A) 2D scanning confocal images of fixed NHEK cells immunostained for FLG (green)) and nucleus (blue) after 24 h of treatment with Ni(NO₃)₂. Scale bar (10 µM). ...

To investigate the effect of Ni²⁺ on the abundance of FLG in the stratified epidermis, we used ex vivo epidermal sheets obtained from skin samples collected from healthy donors (Figures 3B,C). The exposure to Ni²⁺ resulted in pronounced reduction in the abundance of FLG⁺ KHGs compared to the control samples incubated in the absence of Ni²⁺, as seen in confocal microscope 2D- and Z-stack images. Incubation with 1 mM of Ni²⁺ resulted in a complete disappearance of KGHs. Western blot assessment confirmed this reduction at the protein level in the epidermal sheets exposed to the Ni²⁺ salt. The reduction in the signal coming from the proFLG band (the highest band above 400 kDa) was accompanied by disappearance of the signal of lower molecular weight bands and appearance of unusual bands (marked by arrows on Figure 3B); this was confirmed for the epidermis obtained from three different donors.

The observed reduction in the FLG-related signal was due to the protein degradation and not to the mRNA level suppression, as demonstrated by the quantitative real time PCR performed on NHEKs, where we observed FLG mRNA upregulation with the increasing Ni²⁺ concentration (Figure 3D), likely as a compensatory mechanism.

Products of Filaggrin Ni-Hydrolysis Affect Langerhans Cells Activation Profile

Finally, we evaluated the impact of Ni-assisted FLG hydrolysis on the phenotype of antigen presenting cells. Here, monocyte derived Langerhans cells (MDLCs) were used to investigate the activation potential of Ni²⁺ complexed to products of hydrolysis of ex-FLG peptides (HP, Figure 2A). Their full amino acid sequences are presented in Supplementary Table S15. For comparative purposes the CD and UV-vis spectroscopic pH titrations of these complexes were performed in a broad pH range. The spectra and titration curves are presented in Supplementary Figure S7. The complex formation process was monophasic, and spectral parameters could be readily assigned to 4N complexes in all cases. In the CD spectra the alternate pattern of d-d bands was observed, typical for the ATCUN motifs (Ariani et al., 2013). All pK values fall in the range of 5.4–5.8, which corresponds to the conditional dissociation constants at pH 7.4 in the range of 1 to 0.1 µM (Sokolowska et al., 2002). We also calculated ten lowest energy structures for all the HPs. The examples of calculated structures of the complexes with Ni²⁺ are presented in Supplementary Figure S8. In every structural variant, the nickel chelate ring conformations with imposed square planar structure are very similar to each other while the N-terminal and C-terminal parts are much more diverse and adopt many conformations in the simulated structures.

MDLCs were incubated with mixed peptides (HP-02, HP-06, HP-07, HP-12, HP-13, HP-14) and the Ni²⁺-complexes (Ni-HPs) formed in molar nickel excess; NiSO₄ serving as a control; the MDLCs activation was assessed using flow cytometry. In order to gain deeper insights into possible immune pathways that may be affected by the HPs, we also quantified the release of six cytokines from MDLCs, five of which are pro-inflammatory (IFN-α, TNF-α, IL-1β, IL-6, IL-8) and one anti-inflammatory (IL-10). The Luminex assay was performed for HP-06, HP-07 and HP-12.

We noted statistically significant changes between the experimental conditions in results obtained from the same monocyte donor. However, the analysis of pooled results from different biological experiments (between different donors) did not show statistical significance, possibly due to the interindividual variation or relatively low sample number (Supplementary Table S19, S20). Observed trends showed that while the presence of FLG-derived peptides alone did not affect the MDLCs profile in a substantial way, the addition of Ni-HPs resulted in an upregulation of the activation markers (CD86 and HLA-DR) on the cells and a parallel complete loss of the CD80-positivity (Supplementary Figure S9). Ni²⁺ and Ni-HPs conditions are correlated with the increased percentage of CD86⁺ and HLA-DR⁺ cells and loss of the CD80-positivity. As far as the cytokine responses are concerned, we noticed a trend of increased levels of TNFa, IL-6, IL-8 in Ni-HPs in comparison to the nickel only condition (Supplementary Figure S10).

We thank MSc Małgorzata Orłowska for her help with the iTOL visualization.