FAM111A is dispensable for electrolyte homeostasis in mice

Autosomal dominant mutations in FAM111A are causative for Kenny-Caffey syndrome type 2. Patients with Kenny-Caffey syndrome suffer from severe growth retardation, skeletal dysplasia, hypoparathyroidism, hypocalcaemia, hyperphosphataemia and hypomagnesaemia. While recent studies have reported FAM111A to function in antiviral response and DNA replication, its role in regulating electrolyte homeostasis remains unknown. In this study, we assessed the role of FAM111A in the regulation of serum electrolyte balance using a Fam111a knockout (Fam111a−/−) C57BL/6 N mouse model. Fam111a−/− mice displayed normal weight and serum parathyroid hormone (PTH) concentration and exhibited unaltered magnesium, calcium and phosphate levels in serum and 24-hour urine. Expression of calciotropic (including Cabp28k, Trpv5, Klotho and Cyp24a1), magnesiotropic (including Trpm6, Trpm7, Cnnm2 and Cnnm4) and phosphotropic (Slc20a1, Slc20a2, Slc34a1 and Slc34a3) genes in the kidneys, duodenum and colon were not affected by Fam111a depletion. Only Slc34a2 expression was significantly upregulated in the duodenum, but not in the colon. Analysis of femurs showed unaffected bone morphology and density in Fam111a−/− mice. Kidney and parathyroid histology were also normal in Fam111a−/− mice. In conclusion, our study is the first to characterise the function of FAM111A in vivo and we report that mice lacking FAM111A exhibit normal electrolyte homeostasis on a standard diet.

Over the last years, several studies aimed to investigate the function of FAM111A in more detail. FAM111A has been postulated to be a protease expressed in a cell-cycle dependent manner that exhibits autocleavage 18,19 . It plays a role in DNA replication, loading proliferating cell nuclear antigen (PCNA) onto chromatin 19 and preventing fork stalling, potentially through proteolytic removal of protein obstacles from the genomic DNA 18 . While FAM111A is required for DNA replication 19 , its overexpression also reduces DNA replication 19,20 and leads to caspase-dependent programmed cell death in the osteosarcoma cell line U2OS 20 . In addition, FAM111A was identified as a host range restriction factor, meaning that it could prevent viral replication by certain viruses, in particular simian vacuolating virus 40 (SV40), rabbitpox virus and vaccinia virus 21,22 . This effect is achieved through inhibition of viral replication centre formation and viral DNA replication by disrupting the host cell nuclear barrier function 23,24 . Viruses that contain SV40 large T antigen (LT) or serine protease inhibitor 1 (SPI-1), on the other hand, can overcome the host defence through inhibition of FAM111A 21,22 . While we are beginning to understand some of the molecular functions of FAM111A, it remains unclear how these relate to the observed phenotype in KCS2 and GCLEB patients.
Regulation of Ca 2+ and Mg 2+ homeostasis mainly occurs by absorption from the intestine, storage in bone and reabsorption by the kidneys. The thick ascending limb (TAL) and the distal convoluted tubule (DCT) segments of the nephron are important for the renal handling of these ions by expressing specific channels and transporters, www.nature.com/scientificreports/ thereby regulating ion excretion into the urine according to the requirements of the body 25,26 . Disturbed electrolyte homeostasis is an essential feature of KCS2, but the primary cause of the electrolyte disturbances has not been established. Hypoparathyroidism has been proposed as the main cause, as parathyroid tissue was absent in a KCS2 patient 27 . Indeed, primary hypoparathyroidism is in line with patients' increased bone density as well as hypocalcaemia and hypomagnesaemia. However, Isojima and coworkers demonstrated that magnesium supplementation restored parathyroid and calcium levels in a patient, suggesting that hypomagnesaemia is the primary defect 3 . Interestingly, in a disorder called hypomagnesaemia with secondary hypocalcaemia (HSH), caused by mutations in the magnesium channel TRPM6, patients also present with secondary hypoparathyroidism 28 . Of note: hypoparathyroidism only occurs as a consequence of hypomagnesaemia if serum magnesium levels are severely low 29 . A major obstacle to the investigation of FAM111A in electrolyte homeostasis is the absence of a suitable physiological study model, as cell models do not represent the complex interplay between organs or cell types in vivo.
To increase our understanding of the electrolyte disturbances in KCS2 patients, we generated Fam111a knockout (Fam111a −/− ), heterozygous (Fam111a +/− ) and wild type (Fam111a +/+ ) mice. The mice were kept in metabolic cages to collect faeces, blood and urine for biochemical analysis. The parathyroid glands, kidneys and intestines were subjected to immunohistochemistry and expression analyses, and bone structure was analysed by micro-computed tomography (micro-CT).

Results
Mouse breeding. To generate Fam111a +/− mice, the protein coding region of Fam111a was substituted by the ZEN-Ub1 cassette which contains the E. coli lacZ reporter gene and neomycin resistance gene (Fig. 1A). Breeding Fam111a +/− mice resulted in the birth of 43 pups in the ratio Fam111a +/+ (15), Fam111a +/− (20) and Fam111a −/− (8), which was not significantly different from the expected proportion (χ 2 test, p > 0.2). The male to female ratios of the animals used for the study were 4:4, 5:3 and 3:4, respectively. Genotyping confirmed the presence of the cassette in Fam111a −/− and Fam111a +/− mice and the Fam111a allele in Fam111a +/+ and Fam111a +/− animals (Fig. 1B). Deletion of Fam111a was also confirmed at the RNA level by real-time quantitative polymerase chain reaction (RT-qPCR) using RNA isolated from the kidneys, liver, proximal duodenum and distal colon. Fam111a expression was significantly reduced (p < 0.05) by approximately 50% in all tested tissues except the distal part of the colon of Fam111a +/− mice compared to the Fam111a +/+ group. Fam111a expression was absent in the Fam111a −/− mice in all the measured tissues (Fig. 1C).

Fam111a −/− mice do not exhibit growth and electrolyte disturbances. To investigate the effect of
Fam111a deletion on the weight and electrolyte levels in mice, Fam111a +/+ , Fam111a +/− and Fam111a −/− mice were placed in metabolic cages. Fam111a deletion did not affect the weight of the mice significantly compared to their heterozygotes and wild type littermates. There were also no significant differences in food and water consumption, urine volume and faecal weight between the groups (Table 1).

Expression of transporter genes is unchanged in the kidneys of Fam111a −/− mice. The kidneys
play an important role in the regulation of serum electrolyte levels. Haematoxilin and eosin (H&E) staining of the kidney tissue showed normal glomerular and tubular structures (white and black arrowheads, respectively) in Fam111a −/− animals and no fibrosis or infiltration of immune cells was observed ( Fig. 2A,B). To identify potential renal compensatory mechanisms, expression of genes involved in Ca 2+ and Mg 2+ handling was measured in the kidneys. Expression of the magnesiotropic genes including Trpm6, Trpm7, Cnnm2, Slc41a1, Prl1 and Prl2 was not different between the groups (Fig. 2C). The expression of renal Ca 2+ transport genes Trpv5 and Cabp28k was stable upon Fam111a deletion in mice compared to the wild type animals. There were also no significant differences in the expression of renal sodium/phosphate cotransporter genes Slc34a1 and Slc34a3  Fam111a −/− mice exhibit a normal bone phenotype. KCS2 patients have a particular bone phenotype including cortical thickening and medullary stenosis of the long bones. To investigate a potential role of FAM111A in skeletal development, micro-CT was used to characterise structural remodeling of femoral bone in Fam111a mice. Figure 5A shows trabecular parameters of Fam111a +/+ , Fam111a +/− and Fam111a −/− mice. Fam111a deficiency did not affect trabecular bone volume fraction, trabecular thickness, trabecular separation, trabecular number and patterning factor of trabecular region (connectivity of trabeculae), or the structure model index (plate-versus rodlike appearance of trabecular bone). The bone cortical indices were also monitored for structural changes (Fig. 5B). Cortical area, thickness, perimeter and the endocortical bone marrow area were not different in the Fam111a −/− mice. The polar moment of inertia (a proxy for bone strength) was also the same among the groups. Taken together, in contrast to patient findings where increased cortical thickening was described, there are no observable alterations in the bone microarchitecture upon deletion of Fam111a.    athyroidism is a common feature of KCS2, the PTH concentration in the serum of Fam111a −/− animals was similar to that in the Fam111a +/+ group (Fig. 6A). In addition, we performed H&E staining on thyroid sections of Fam111a +/+ and Fam111a −/− mice. This demonstrated that Fam111a −/− mice had parathyroid glands of normal cellular constitution, consisting primarily of chief cells (Fig. 6B, C). We did not detect any damage such as cysts or immune cell infiltration to the parathyroid glands in Fam111a −/− mice.

Discussion
In this study, we investigated whether FAM111A plays a role in growth and electrolyte homeostasis to help explain the KCS2 patient phenotype. For this, we used a knockout mouse model. Fam111a −/− mice showed normal body weight, bone morphology, electrolyte homeostasis and PTH levels. Based on our results, we conclude that FAM111A is not essential in regulating growth and electrolyte homeostasis in mice on a standard diet. Expression analysis demonstrated that duodenal Slc34a2 transcripts were significantly upregulated in Fam111a −/− mice, but the contribution to total phosphate absorption of this increased expression is unclear as Slc34a2 dependent phosphate absorption in mice mainly occurs in the ileum 30 . RNAseq of the kidney did not identify additional ion channels and transporters differentially regulated in Fam111a −/− mice. While there were no changes in the mRNA expression of ion channels and transporters in this study, our data does not exclude the regulation of channel activity at the level of protein expression and/or post-translational regulation. The expression of Glyat and Gm4952 were significantly downregulated. However, these genes are located in the vicinity of Fam111a on the genome. It is possible that there are enhancers present within the Fam111a gene that affect the expression of these genes. Mid-1 and Slc47a1 were upregulated, however, this change is of unknown significance www.nature.com/scientificreports/ to the KCS2 phenotype. A recent study using small sample single cell RNA sequencing (scRNAseq) of mouse kidney tissues identified a cluster of early DCT-like cells with high enrichment of cell cycle and replication factors suggesting a proliferative potential in these DCT cells 31 . However, total DCT area as well as total early DCT area was similar in the Fam111a +/+ and Fam111a −/− mice. Several explanations can contribute to the absence of the patient phenotype in Fam111a −/− mice, including: (1) the nature of FAM111A mutations (loss of function vs gain of function), (2) compensatory mechanisms for Fam111a deletion, (3) differences between mouse and man. We will elaborate on these points in the following sections.
The exact consequences of the nature of the FAM111A patient mutations on the protein function are unclear: specifically, whether it leads to a gain or loss of protein function. Indeed, the presence of mutations in the FAM111A catalytic site in some KCS2 patients are in line with a loss of function effect 8,32 . However, recent in vitro studies have pointed towards a gain of function effect on the protease activity of FAM111A 18,20 . On the other hand, because the FAM111A protease activity can result in autocleavage, it has been also hypothesized that activating mutations in FAM111A eventually will result in loss of function by enhanced autocleavage 18 . The present study seems to support the viewpoint that a gain of function rather than a loss of function of FAM111A is responsible for the observed pathology, as we did not observe an electrolyte phenotype in the knockout mice. Mouse models of electrolyte disorders due to hyperactivating mutations such as autosomal dominant hypocalcemia (ADH) and familial hyperkalemic hypertension are generated by knock in of the relevant mutation in mice [33][34][35][36][37] . Similarly, transgenic knock in mouse models expressing the FAM111A patient mutations may better reflect the phenotype seen in patients with KCS2.
Based on our study, we cannot exclude that the function of FAM111A is rescued by compensatory mechanisms. FAM111A has been recently shown to function as a DNA-protein crosslinking (DPC) protease 18 . Spartan (SPRTN), the most well characterised DPC protease, possesses some overlapping functions with FAM111A in preventing replication fork obstacles by Top1-cleavage complexes (Top1cc) and formaldehyde 18,38 . Loss of www.nature.com/scientificreports/ function mutations in spartan result in Ruijs-Aalfs syndrome (RJALS) with features such as premature aging, cataracts, stunted growth, skeletal deformities and liver cancer 39 . However, unlike Fam111a −/− mice, complete deletion of spartan in mouse models is embryonically lethal while hypomorphic mice develop symptoms of RJALS 40 . It is thus possible that unidentified proteases may compensate for the loss of FAM111A in mice. Electrolyte imbalance is a prominent feature of Kenny-Caffey syndrome 2 and knockout mouse models have been employed extensively in elucidating the transport function of genes that encode transport proteins [41][42][43][44] . Also, knockout mouse models have been useful in identifying electrolyte related function of genes that do not encode for transporters (indirect regulation). For example, the roles of Adenylate cyclase 3 (Adcy3) and calcineurin in regulating Mg 2+ handling were described using knockout mouse models 45,46 . However, not all knockout mouse models recapitulate the phenotype associated with the human disease. For example, Fxyd2 knockout mice have a normal serum Mg 2+ concentration, whereas loss-of-function mutations in FXYD2 are responsible for isolated dominant hypomagnesemia in patients 47,48 . In addition, Trpm6 knockout mice did not display hypomagnesaemia as is the case in the disease hypomagnesaemia with secondary hypocalcaemia (HSH) caused by mutations in Trpm6. Similar observations have been made in mice deficient in 49 Cnnm2, suggesting that www.nature.com/scientificreports/ magnesium homeostasis in mice may be differently regulated compared to man. Mice are also commonly used to study parathyroid function, but Günther et al. found that glial cells missing homolog 2 (Gcm2) deficient mice and parathyroidectomised mice had normal serum PTH levels, as the thymus provided a backup mechanism of PTH secretion 50 . Indeed, mice harbouring a point mutation in Tbce had no parathyroid glands, but normal PTH levels 51 . Therefore, mice may not be a suitable model for studying the role of FAM111A in electrolyte handling. FAM111A has been described as a DNA replication stress response protein in in vitro studies by preventing replication fork stalling caused by poly-ADP-ribose polymerase 1 (PARP1) accumulation as well as by responding to viral stress, preventing viral DNA replication 18,22,23 . Deletion of FAM111A in HAP1 cells did not affect cell viability and proliferation unless they were treated with DNA-damage agents 18 . It would be interesting to see how Fam111a −/− mice would respond to stress conditions where FAM111A is known to play a role such as response to viral stress (LT-virus and orthopoxvirus) or DNA replication stress induced by PARP1 and/or topoisomerase 1 (TOP1) inhibitors. In addition, dietary restriction of electrolytes could be a trigger to develop electrolyte related symptoms. An example was described in serum-and glucocorticoid-regulated kinase 1 (SGK1) knockout mice which displayed subtle changes in electrolyte regulation compared to control mice in normal conditions and challenging them with dietary NaCl restriction impaired renal sodium handling 52 . Finally, DCT cells are known to structurally adapt to different stimuli, such as furosemide and low sodium or potassium diets [53][54][55] . Investigation of how these mice will respond to these stimuli may explain a potential role of Fam111a in DCT replication and remodelling.
In conclusion, our data show that Fam111a −/− mice display normal electrolyte homeostasis, PTH levels, growth, and skeletal development. Expression of relevant electrolyte transporters was not altered in the kidneys and intestines. Hence, knockout of FAM111A does not affect renal electrolyte handling in mice on a standard diet.

Methods
Generation of Fam111a −/− mice. The C57BL/6 N Fam111a mouse strains used for this research project were created from embryonic stem (ES) cell clone 19174A-G2, 19174A-A1 generated by Regeneron Pharmaceuticals, Inc. This ES cell clone was subsequently used by the KOMP Repository (https:// www. komp. org/) and the Mouse Biology Program (www. mouse biolo gy. org) at the University of California Davis to produce chimera mice. Methods used to create the Velocigene-targeted alleles have been reported previously 56 . The non-conditional allele Fam111a tm1(KOMP)Vlcg was generated by insertion of ZEN-UB1 cassette targeting chromosome 19 in ES cells, which results in the deletion of 4,553 nucleotides (12,584,048-12,588,600; GRCm38.p6) corresponding to the entire protein coding region of Fam111a. ES cell clones were selected based on PCR genotyping and then blastocysts were microinjected into C57BL/6 N background mice. Balb/c donors were used for microinjection www.nature.com/scientificreports/ to produce the chimera. Sperm rederivation was performed to produce Fam111a +/− animals. Fam111a −/− and Fam111a +/+ mice were generated by breeding Fam111a +/− mice. The following primer pair was used for genotyping the wild type (Fam111a +/+ ) allele: forward primer (5′-AGC CAA AAT GAG CTG TAA GAAGC-3′) and reverse primer (5′-GTG TCA TTG TCC CAT TCA AC-3′). The following primers were used for genotyping the presence of the ZEN-UB1 cassette: forward primer (5′-TCA TTC TCA GTA  TTG TTT TGCC-3′) and reverse primer (5′-TGA ACT TTC TCA CAT CGT AGC-3′).

Ethics and approval.
Animal studies were carried out at the Radboud Institute for Molecular Life Sciences, Nijmegen, The Netherlands. All methods and protocols performed in this study were approved by the local Ethics committee of the Radboud University Nijmegen (RU DEC 2015-0112-006) and the national Ethics committee of the Dutch Central Commission for Animal Experiments (AVD103002016382). This study was performed in compliance with the ARRIVE guidelines. The methods were in accordance with relevant guidelines/regulations, as follows. Fam111a +/+ mice were used as controls and individual mice were considered experimental units. Sample size was calculated as n = 7 per group based on the outcome measurement of serum magnesium, using a Java Applet 57 . We did not make use of any exclusion criteria and no animals were excluded, but one heterozygous animal died before the study. Randomisation was applied in housing and sample collection. Blinding was performed by giving all animals a unique number, so that the genotypes were not known to the researchers. All genotypes were randomly divided over the cages using the website https:// www. random. org/ seque nces/. Sampling was done per cage. Researchers were blinded to the genotype until the first analyses of results. Mice were housed with maximally 6 animals in standard cages (mouse Eurostandard type IIL) and during the experiments they were individually housed for 48 h in metabolic cages (24 h of acclimatisation, 24 h for actual experiments), after which they were placed back into their own cages. Cage enrichment (bedding, nesting material and igloo) was used in the standard cages, but not the metabolic cages. Specific humane endpoints were defined as follows: (1) The animal has visible signs of discomfort, such as weight loss (more than 15% of the initial weight), hunchback, bad fur and/or bad movement pattern, (2) The animal loses more than 20% weight during housing in a metabolic cage, (3) Specific for the low serum magnesium value: the animal has visible muscle cramps, the animal shows lack of movement, the animal shows a severe abnormal walking pattern (wobbling), the animal has severe exorotation of the heel (indication of muscle weakness) or the animal is not able to hold on to the cage (indication of muscle weakness). These endpoints were monitored daily through weighing and visual monitoring.
Metabolic cages and tissue sampling. Mice (8-10 weeks old) were placed in metabolic cages and left to acclimatise for 24 h. Mice were kept on a normal diet, containing 1% calcium (w/w), 0.22% magnesium (w/w), 0.7% phosphate (w/w) and 1000 IU/kg vitamin D 3 . During the next 24 h, the mice were weighed and urine and faeces were collected. In addition, water and food consumption were measured. The following day, the mice were anaesthetised by 4% (v/v) isoflurane inhalation, serum/blood was collected and the mice were sacrificed. Kidney, thyroid glands, duodenum and colon were collected and snap frozen in liquid nitrogen and stored at -80ºC until further analysis.
Serum electrolyte and PTH concentrations. Serum and urine PO 4 3− , Na + and K + were determined by the clinical laboratory of the Radboud University Medical Center using an automated system (Abbott Diagnostics, Hoofddorp, The Netherlands). Mg 2+ concentrations were determined by xylidil blue colorimetric assay kit (Roche/Hitachi, Tokyo, Japan) and read at a wavelength of 600 nm on a BioRad plate reader (BioRad, California, USA). Serum and urinary Ca 2+ concentrations were measured using a chromogen-based colorimetric assay and read at a wavelength of 570 nm (Sigma Aldrich, Zwijndrecht, The Netherlands). The values obtained were calibrated using a Precinorm standard solution (Precinorm U, Roche, Switzerland). Serum albumin was measured using the BCG albumin assay kit (Sigma-Aldrich, Saint Louis, USA) according to the manufacturer's recommendation and read at a wavelength of 620 nm on a BioRad plate reader (BioRad, California, USA). PTH serum concentrations were measured using the mouse PTH 1-84 ELISA kit (cat. No 60-2305, Quidel, CA, USA) according to the manufacturer's instructions.

Histology (haematoxylin & eosin).
Tissues were fixed for at least 24 h in 4% (v/v) formalin. Samples were dehydrated in ethanol, embedded in paraffin and cut into sections of 5 μm. The sections were deparaffinised in xylene for 10 min and then rehydrated in graded concentrations of ethanol (100%, 95%, 75%, 50% v/v). The sections were incubated 5 min in haematoxylin (Merck, Amsterdam, Netherlands) and differentiated in tap water after which the sections were stained with eosin (Merck) and dehydrated in in graded concentrations of ethanol (50%, 70%, 96%, 100% (v/v)). Slides were cleared in xylene and mounted using Pertex mounting medium (VWR, Amsterdam, The Netherlands). Tissue slices were imaged using a Zeiss Axiocam 503 color camera mounted onto a Zeiss Axio Imager M2. Total 58 . The reference transcriptome was downloaded from the BioMart repository at Ensembl (Cambridge, UK). The raw and normalized count data were further analysed using R (Vienna, Austria). Raw and processed data files are available in the Gene Expression Omnibus database (GSE196466).

Micro-CT.
Femurs were dissected and fixed in 4% (v/v) formalin for 24 h and subsequently kept in 70% (v/v) ethanol. Femurs were scanned using the Skyscan 1076 in vivo X-ray computed tomographer (Bruker microCT, Kontich, Belgium) with a voxel size of 8.88 μm and an exposure time of 2300 ms as described previously 49 .
Briefly, X-ray power was set at 40 kV and tube current at 250 mA. Beam hardening (20%) was reduced using a 1 mm aluminum filter, ring-artefacts was set at 5 and an average of three photos (frame averaging) at each angle (0.8°) were taken to generate the final images. Cortical bone parameters were performed in the diaphyseal cortex which included a scan area of 0.45 mm in the femoral center. Trabecular bone measurements were obtained by scanning the distal metaphysis, a total scan area of 1.35 mm from the distal growth plate to the femoral center.
Immunohistochemistry. Immunohistochemistry was performed as previously described 59 . Briefly, costaining for parv and NCC were performed on 4 μm sections of formalin fixed and paraffin embedded kidney samples. Sections were incubated with blocking buffer (TSA fluorescence system, Perkin Elmer) to prevent unspecific antibody binding. Sections were incubated overnight at 4 °C in rabbit anti-parvalbumin (1:200, Swant, Bio Connect, Switzerland) and sheep anti-NCC (1:200, MRC) in blocking buffer. For detection, sections were incubated with Alexa Fluor-conjugated secondary antibodies (Thermo Scientific, Amsterdam, The Netherlands). Cell nuclei were stained with DAPI (1:20,000) for 10 min at room temperature. Images were taken with an AxioObserver camera and stitched with AxioVision software (Zeiss, Sliedrecht, The Netherlands). NCC or parv positive regions of interest were obtained by thresholding and total positive area per kidney slice was measured using the analyse particle function in Fiji 60 .
Statistical analysis. Data are expressed as mean ± SD. Analysis was done using Graphpad Prism v8 (San Diego, CA, USA). For statistical analysis of three groups, the assumption of equal variances was tested using the Brown-Forsythe test and normality was tested using the Shapiro-Wilk test. For two groups, the assumption of equal variances was tested using an F test. If assumptions were not met, data was first logarithmithically transformed to achieve equal normality. Then, for three groups one-way ANOVA was used to test for differences in the data, followed by a Tukey's multiple comparison test in case the null hypothesis was rejected. In the case of two groups, an unpaired t-test was used. Differences with a p value of < 0.05 were considered statistically significant. For the RNAseq, additionally, a false discovery rate (FDR) ≤ 0.05 was used to determine statistical significance.

Data availability
Raw data are available from the corresponding author, J.H.F.B., upon reasonable request. Raw and processed RNAseq data files are available in the Gene Expression Omnibus database (GSE196466).