The highly conserved FOXJ1 target CFAP161 is dispensable for motile ciliary function in mouse and Xenopus

Cilia are protrusions of the cell surface and composed of hundreds of proteins many of which are evolutionary and functionally well conserved. In cells assembling motile cilia the expression of numerous ciliary components is under the control of the transcription factor FOXJ1. Here, we analyse the evolutionary conserved FOXJ1 target CFAP161 in Xenopus and mouse. In both species Cfap161 expression correlates with the presence of motile cilia and depends on FOXJ1. Tagged CFAP161 localises to the basal bodies of multiciliated cells of the Xenopus larval epidermis, and in mice CFAP161 protein localises to the axoneme. Surprisingly, disruption of the Cfap161 gene in both species did not lead to motile cilia-related phenotypes, which contrasts with the conserved expression in cells carrying motile cilia and high sequence conservation. In mice mutation of Cfap161 stabilised the mutant mRNA making genetic compensation triggered by mRNA decay unlikely. However, genes related to microtubules and cilia, microtubule motor activity and inner dyneins were dysregulated, which might buffer the Cfap161 mutation.

www.nature.com/scientificreports/ of eggs into the ampulla and towards the uterus 14 , respectively. Wave-like movement of the sperm flagellum, a specialised long cilium, is essential for sperm motility and fertilisation 4 . Defects in the formation of cilia or their function cause human diseases collectively known as ciliopathies (reviewed in [15][16][17][18][19]. A specific subgroup of ciliopathies, referred to as primary cilia dyskinesia (PCD), is caused by impaired function of motile cilia. Reflecting the functions of motile cilia, PCD is characterised by situs randomisation, impaired mucociliary clearance, respiratory problems, and male infertility. Mouse models of PCD frequently show reduced female fertility and hydrocephalus 12 , which are less common in human patients (reviewed in 20,21 ).
The formation of apparently all motile cilia in the mouse as well as in other vertebrates and invertebrates is under the control of the transcription factor FOXJ1 [22][23][24][25][26][27][28] . Thus, genes that act downstream of FOXJ1 are likely to be required for the formation or function of motile cilia. In microarray screens for FOXJ1 target genes in nodestage embryos and foetal lungs 29 we have identified a number of new Foxj1 targets [30][31][32][33][34] including Cfap161, the orthologue of the axonemal Chlamydomonas reinhardtii FAP161 protein 35 . Recently, single-particle cryo-electron microscopy of Chlamydomonas flagella showed that FAP161 is one of the 33 microtubule inner proteins (MIPs) and likely localises to the A-tubule of the outer doublets 36 . Morpholino-mediated knock down of Cfap161 in zebrafish led to the loss of outer dynein arms, reduced beating frequency of pronephric cilia and strong ciliopathy phenotypes including left-right asymmetry defects in one study 37 , but only a curved body axis and hydrocephalus in a subsequent analysis 38 leaving some uncertainties concerning the ciliary function of CFAP161. Human CFAP161 (c15orf26) is located on chromosome 15q in the linkage region of Kartagener syndrome 39 , suggesting a potential involvement of Cfap161 mutations in this subtype of PCD. Here, we describe the analysis of CFAP161 function in Xenopus crispants and mutant mice generated by homologous recombination. Surprisingly, disruption of the highly conserved Cfap161 gene did not lead to obvious phenotypes related to dysfunctional motile cilia.

Results
Expression of Cfap161 in mouse and Xenopus. Mouse Cfap161 (RefSeq NM_029335.3) encodes an evolutionary conserved (Table S1) 303 amino acids (aa) protein lacking domains of known biochemical function. High expression levels of Cfap161 correlate in general with the presence of motile cilia and co-expression of Foxj1 (Fig. 1A). In early embryos (E7.75) Cfap161 is expressed in the ciliated ventral layer of the embryonic node (the mouse left-right organiser, LRO; Fig. 1Ba,a′), and in later foetal stages (E16.5) it is expressed in ependymal cells of the brain (Fig. 1Bb), epithelial cells lining the eustachian tube (Fig. 1Bc), the respiratory epithelium of the nasal cavity (Fig. 1Bd), and in ciliated epithelial cells of the lung (Fig. 1Be). In these tissues expression was also detected in adults (Fig. 1A) as well as in cells of the oviduct carrying motile cilia, and in testis (Fig. 1A,Cae). In testis, Cfap161 transcripts were not found in all seminiferous tubules (Fig. 1Ce,e′) suggesting that Cfap161 transcription is restricted to distinct stages of the cycle of the epithelium of the seminiferous tubules. Consistent with this idea, CFAP161 protein was only observed at late stages of spermiogenesis (Fig. S1). Cfap161 mRNA was additionally detected in some cells lacking motile cilia: in the retina, in the ganglion cell layer (GCL), the inner nuclear layer (INL), in the photoreceptor cells (PRL; Fig. 1Cf, red arrowhead in f′) and in developing follicles (red arrowhead in Fig. 1Cd). Likewise, CFAP161 protein was detected in cells harbouring immotile cilia (primary cilia), such as hair cells in the inner ear or in cells of the kidney collecting ducts (Fig. S2).

Localisation of CFAP161 and dependence on FOXJ1.
To study CFAP161 protein expression in mouse we generated a monoclonal rat antibody directed against a peptide (α-pI) mainly encoded by exon 1 ( Fig. 2A and Fig. S3) and polyclonal antibodies directed against a peptide (α-pII) encoded by exon 2 in rabbits ( Fig. 2A). In Western blots both antibodies detected Flag-tagged CFAP161 over-expressed in CHO cells as well as endogenous CFAP161 in testis lysates (Fig. 2B). Additionally, these antibodies detected CFAP161 expression concomitantly with cilia formation indicated by co-expression with the ciliary protein IFT88 in air-liquid interface (ALI) cultures of mouse tracheal epithelial cells (mTECs) (Fig. 2C). The monoclonal antibody detected endogenous CFAP161 in foetal (E17.5) tissue sections (Fig. 2D), consistent with Cfap161 mRNA expression domains (Fig. 1Bb-e) corroborating the specific expression of Cfap161 in cells carrying motile cilia. CFAP161 was not detected in tissues of Foxj1 null mutant (Foxj1 lacZ/lacZ ) foetuses (n = 4; Fig. 2De-h). Likewise, Xenopus foxj1-crispants revealed a global downregulation of cfap161 as prominently seen in the epidermis (Fig. 2Eb,b′) where expression of cfap161 is nearly absent in the MCCs, and a unilateral foxj1 gain of function (foxj1-GOF) strongly induced transcription of cfap161 (Fig. 2F) confirming the dependence of Cfap161 expression on FOXJ1 as was previously observed in zebrafish 37 .
Indirect immunofluorescence of adult tissue sections using α-pI and α-pII antibodies showed co-localisation of CFAP161 with acetylated-α-tubulin (ac-TUB; Fig. 3A), indicating that CFAP161 is a component of motile cilia. CFAP161 and ac-TUB staining did not overlap in the distal part of cilia ( Fig. 3Aa′-d′, arrowheads in insets) indicating that CFAP161 is excluded from the tip of the cilium, which was most clearly seen in the sections of the nasal respiratory epithelium (Fig. 3Ac′,d′). Absence of CFAP161 from the cilium tip is consistent with the localisation of its Chlamydomonas orthologue in outer microtubule doublets and likely reflects the fact that outer doublets terminate earlier than the central inner pair 40 . Consistent with the ciliary localisation, CFAP161 was found in the flagellum of spermatozoa (Fig. 3Ae,e′; Fig. S1). Detection of Xenopus CFAP161 failed with α-pI www.nature.com/scientificreports/ and α-pII antibodies but heterologously expressed murine CFAP161 tagged with a N-terminal EGFP (GFP-CFAP161) partially co-localised with a Cetn4-RFP fusion construct, which marks the basal bodies of epidermal MCCs (Fig. 3Ba). Of note, most GFP-CFAP161 signal was in anterior juxtaposition underneath each basal body (Fig. 3Ba′,a″), reminiscent of the rootlet (Fig. 3Bb). Localisation at basal bodies might reflect non-physiological accumulation due to high levels of the overexpressed GFP-tagged CFAP161. Similarly, also overexpressed N-or C-terminally HA-tagged Xenopus Cfap161 was not specifically localised in cilia but found throughout the whole cell (Fig. S4). It appears unlikely that in Xenopus this conserved ciliary protein is absent from motile cilia. Rather, the apparent uniform distribution of HA-tagged Cfap161 might reflect high levels of overexpressed HA-tagged Cfap161 that exceed the amount that can translocate to cilia. To obtain first insights into the biochemical context in which CFAP161 might act, we identified potential interaction partners by mass spectrometry of CFAP161 complexes immunoprecipitated from wild type testis www.nature.com/scientificreports/ lysates using the polyclonal α-pII antibody (Table S2). One potential interaction partner was KIAA0556 (also known as Katanin-interacting protein KATNIP), a basal body protein that stabilises cytoplasmic microtubules in human cells, regulates ciliary A-tubule number in C. elegans, and when mutated, causes Joubert syndrome in humans 41 . Yeast-two-hybrid analysis (Fig. S5) validated a robust interaction of CFAP161 and KIAA0556, supporting a function of CFAP161 in some aspects of microtubule organisation or function.

Functional analyses of CFAP161 in Xenopus.
To investigate the function of CFAP161 in the frog Xenopus we mutated cfap161 by genome editing using CRISPR/Cas9. Two different single guide RNAs (sgRNAs) were designed, targeting exon 1 (sgRNA1) and exon 3 (sgRNA2), respectively. Sequencing of PCR products with pooled DNAs of 10 F0 crispants (Fig. S6A,B) confirmed successful genome editing of cfap161. Furthermore, cfap161 transcripts were strongly reduced in cfap161-crispants, indicative for nonsense mediated mRNA decay (Fig. S6Cb,b′). Initial experiments with individual sgRNAs showed no effect in ciliated cells of the larval epidermis (data not shown). Subsequently both sgRNAs (sgRNA1 & sgRNA2) were injected simultaneously. Crispants were analysed for laterality defects, the development of pronephric cysts or an externally visible hydrocephalus in stage 45 tadpoles. Evaluation of 297 crispants obtained in five independent experiments showed no impact on their overall phenotype ( Fig. 4A-D). To address whether loss of cfap161 has a more subtle effect on ciliary beating of epidermal MCCs, we analysed high speed video microscopy recordings of stage 32 wild type and crispant  Table S3) nor CGF (around 400 µm/s for both wild type and crispant embryos; Fig. 4Fa,b; Table S4) was affected by mutating cfap161. Thus, disruption of cfap161 showed no phenotypic changes associated with dysfunctional motile cilia in Xenopus.

Functional analysis of CFAP161 in mice.
To analyse CFAP161 function under physiological conditions in mice we generated a conditional allele by flanking Cfap161 exons 2 and 3 by loxP recombination sites (Fig. 5A). Deletion of these two exons by Cre-mediated site-specific recombination results in a frame shift that generates a premature translational stop codon by the first base triplet of exon 4, which should effectively abolish the generation of full-length CFAP161 protein. To generate mice lacking CFAP161 in all tissues we deleted exons 2 and 3 (Cfap161 ∆ex2,3 ) in the female germ line using ZP3:Cre mice 42 . Homozygous Cfap161 ∆ex2,3 mice were born at Mendelian ratio (wt 95, het 186, hom 91; χ 2 = 0.086, p = 0.98) and showed no obvious abnormalities. Western blot analyses of testis and epididymis lysates with the monoclonal antibody α-pI (which recognises the epitope encoded by exon 1; analysis shown in Fig. S3) and with polyclonal antibody α-pII did not detect full length CFAP161 protein or any other shorter protein product ( Fig. 5B; Fig. S9C) that might have been generated by alternative splicing. Likewise, our antibodies did not detect CFAP161 protein in mutant testis sections (Fig. 5Cb,d) supporting that deletion of exons 2 and 3 abolishes production of CFAP161 and thus leads to a functional null allele. Disruption of normal motile cilia function can lead to hydrocephalus and mucus accumulation in the respiratory tract, abnormal situs of visceral organs and male infertility due to immotile spermatozoa, rarely female www.nature.com/scientificreports/   Fig. 6A) of 4 months old Cfap161 ∆ex2,3 homozygotes (n = 4) did not show signs of enlarged ventricles (Fig. 6Cb,b′,d,d′,d″). Likewise, sections of mutant lungs (n = 4; Fig. 6Cf,f′) or PAS stained serial sections (section planes are shown in Fig. 6B) of the mutant nasal cavities (Fig. 6Ch,h′,j) did not show obvious morphological alterations or accumulation of mucus. Although mutant males bred apparently normal computer-assisted sperm analysis (CASA) was performed to detect potential abnormalities in sperm number or motility but revealed no differences between wild type and mutant ( Fig. S7A; Table S5) samples. Likewise, electron microscopic analyses of mutant sperm cells prepared from cauda epididymis showed no obvious structural abnormalities (Fig. S7B) in the axonemes of flagella and cilia of the lung (Fig. S7C). Additionally, no polydactyly or deafness as a sign for impairment of primary cilia were observed in the mutants (n = 31). Taken  RNA-seq on CFAP161-deficient testis reveals dysregulation of ciliary gene expression. Given the unexpected result that the mutation of an evolutionary highly conserved FOXJ1 target gene has no motile cilia-related phenotype, we analysed transcriptional changes in control wild type and Cfap161 ∆ex2,3/∆ex2,3 testes by mRNA-sequencing (RNA-seq). Differential expression analysis using DEseq2 revealed a number of statistically significantly dysregulated transcripts (338; P-adj < 0.05) (Table S6). Gene-ontology (GO) analysis 43,44 of these transcripts showed a significant enrichment of biological process terms relating to microtubule and cilia, including mitotic spindle organisation, protein localisation to cytoskeleton, microtubule bundle formation, cilium movement, and cytoskeleton-dependent intracellular transport (Table S7). Additionally, terms related to RNA biology (e.g. mRNA stabilisation, regulation of RNA processing, regulation of RNA splicing) as well as to chromatin regulation (e.g. chromatin remodelling, DNA packaging, and histone modification) were www.nature.com/scientificreports/ enriched. The most enriched GO-terms were ATP-dependent microtubule motor activity (minus-end-directed) and inner dynein arm when the dataset was compared to GO molecular function and GO cellular component datasets, respectively. Of the 338 dysregulated genes, 164 were downregulated and 174 were upregulated, but only few transcripts showed strong changes in expression levels (3 transcripts showed downregulation to levels of 0.5-fold-expression or smaller, and 6 showed upregulation of 2-fold-expression or higher, including Cfap161) (Table S8).

Discussion
In this analysis we show that strong Cfap161 expression correlates well with the presence of motile cilia, consistent with its dependence on FOXJ1 both in Xenopus and mice. In mice CFAP161 protein localises to the axoneme, consistent with localisation of its Chlamydomonas orthologue FAP161 as a microtubular inner protein of the A-tubule of outer doublet microtubules. In both analysed vertebrate species, the disruption of Cfap161 did not result in phenotypes related to dysfunctional motile cilia. Comparison of mRNA from wild type and Cfap161-mutant mouse testis showed stabilisation or enhanced transcription of Cfap161 mRNA in mutants and dysregulation of more than 300 genes including genes related to microtubules and cilia, microtubule motor activity and inner dyneins. The absence of phenotypes related to dysfunctional motile cilia in both Xenopus and mouse came as a surprise given that morpholino-mediated knock down of cfap161 (previously referred to as C18H15orf26) caused strong 37 and mild 38 ciliopathy phenotypes in zebrafish. Also, the interaction of CFAP161 with KIAA0556, a basal body and microtubule-associated protein which stabilises cytoplasmic microtubules in human cells and regulates ciliary A-tubule number in C. elegans, supported a ciliary function of CFAP161, although a truncation of the KIAA0556 protein was not associated with defective cilia structure or motility in ventricular ependymal cells in mice as well 41 . One potential explanation for the lack of Cfap161 mutant phenotypes in Xenopus and mouse could be the presence of redundant genes. However, we did not find any evidence for the presence of Cfap161 paralogs in both species. To investigate the function of cfap161 in the frog Xenopus, we initially also used Morpholino oligomers (MOs) that interfere with either mRNA translation or splicing (TBMO or SBMO). However, only high MO doses resulted repeatedly in specimens with enlarged brain ventricles, oedema and compromised ciliation of epidermal MCCs, and we were unable to rescue any of these phenotypes with a tagged murine or untagged Xenopus cfap161 mRNA (data not shown). Because these phenotypes were not observed after disruption of cfap161 by CRISPR/ Cas9 we attribute these phenotypes to unspecific effects of the used morpholinos at high concentrations. A ciliary function of cfap161 in zebrafish that has been lost in Xenopus and mice cannot be ruled out at present, however it seems unlikely. Thus, it is possible that the discordant observations after morpholino knock down of cfap161 in zebrafish 37,38 reflect unspecific effects similar to those we observed in Xenopus.
Genetic compensation of mutations is increasingly observed in various species (reviewed in 45 ). Xenopus cfap161-crispants showed downregulation of cfap161 mRNA (Fig. S6Cb,b′) suggesting nonsense-mediated decay. mRNA degradation is known to trigger genetic compensatory mechanisms in zebrafish embryos and mouse cells 46 . Thus, in principle, the absence of a motile cilia related phenotype in Xenopus crispants might be the result of genetic compensation triggered by cfap161 mRNA decay. However, the absence of a specific effect in cfap161 morphants argues against this possibility. Also, our RNA-seq data from the mouse do not support the action of a compensatory mechanism triggered by nonsense-mediated mRNA decay because Cfap161 mRNA was stabilised rather than destabilised in mutants. Thus, if genetic compensation underlies the absence of a phenotype in Cfap161-mutant mice the mechanism cannot depend on mRNA degradation 46 . mRNAs bearing a premature termination codon as is the case in our Cfap161 ∆ex2,3 allele can trigger genetic compensation by promoting transcription of homologous genes 47 . However, as stated before, no Cfap161 paralogs were found in the mouse genome making this possibility unlikely.
The high number of statistically significantly dysregulated transcripts in Cfap161-mutant testis resulting from mutation of a gene without known transcriptional activity is surprising as is the abundance of these dysregulated genes on chromosome 7, where also Cfap161 is located. Analysis of gene ontology and GO molecular function and GO cellular component datasets suggested a potential effect of CFAP161 loss of function on the expression of other genes including genes related to microtubules and cilia, microtubule motor activity and inner dyneins. A conclusive interpretation of the RNA-seq results seems difficult as to the precise mechanism of why expression of so many genes is affected and how that might affect the loss of Cfap161 phenotype and potentially buffer the Cfap161 mutation.
While protein complexes bound to the inner surface of doublet microtubules were identified in diverse species (reviewed in 48 ) not much is known about the biological functions of these proteins. In Tetrahymena RIB72A and the conserved RIB72B are MIPs that bind to the inner surface of A-tubules. Deletion of either protein caused disruption of distinct MIP complexes, severe structural defects of the A-tubule and abnormal ciliary beating indicating MIP functions in stabilisation of doublet microtubules and possibly regulation of ciliary beating 49 . In contrast, single deletion of the conserved B-tubule MIPs FAP45 and FAP52 had a slight or no effect, respectively, on swimming velocity and ciliary beat frequency in Chlamydomonas, whereas only deletion of both MIPs significantly reduced both parameters and destabilised B-microtubules, which was suggested to reflect a "fail-safe" mechanism acting in the stabilisation of doublet microtubules 50 . Such a "fail-safe" mechanism could possibly also explain the apparent absence of a ciliary phenotype in Cfap161 mutants. We cannot rule out that disruption of Cfap161 causes subtle defects of stability or function of motile cilia. Such defects, if present, might sensitise cilia for other mutations or only manifest themselves under other conditions impinging on ciliary function. Collection, embedding, and sectioning of mouse tissues. Mice were killed by cervical dislocation, tissues dissected, fixed overnight at 4 °C in 4% PFA, 100% methanol or 100% methanol/DMSO (4:1), if necessary decalcified in 0.5 M EDTA for 2 weeks exchanging EDTA every other day, dehydrated, embedded in paraffin according to standard procedures and sectioned at 5 or 10 µm. For standard histology the lungs were not inflated, but wt and mutant were equally treated. Mice used for transmission electron microscopy (TEM) were perfused with fixative followed by dissection of the organs and embedded and analysed as described 52 . See "Supplemental Material and Methods" for details.

Scientific Reports
Section in situ hybridisations (SISH). SISH were performed on 10 µm sections of PFA fixed and paraffin-embedded tissues as described in 53 . DIG-labelled RNA probe was produced from FANTOM plasmid ZX00121F10 54 using the Roche DIG RNA labelling system.
Whole mount in situ hybridisation (WISH). WISH were performed on E7.5-8.0 old mouse embryos as described 29 .
Histological staining. 5 µm sections of PFA fixed and paraffin-embedded tissues were stained with haematoxylin and eosin (HE) according to standard procedures, or with PAS using the Periodic Acid-Schiff Kit (Sigma Aldrich) according to manufacturer's instructions.
Cell culture. Murine tracheal epithelial cells (mTECs) were isolated and cultured at air liquid interface (ALI) as described 55  RNA isolation and RNA-seq. Total RNA was isolated from single testis without epididymis from 3 different 11-17 weeks old wild type and 3 different 11-13 weeks Cfap161 ∆ex2,3/∆ex2,3 males using the Direct-zol RNA Mini-Prep Kit (ZYMO) according to manufacturer's instructions. RNA integrity was first validated by gel electrophoresis, then further analysed by Eukaryote Total RNA Nano and Bioanalyser. Details concerning library generation, quality control and quantification as well as details for the sequencing run and raw data processing can be found in the "Supplementary Materials and Methods". www.nature.com/scientificreports/ Generation of antibodies. Monoclonal antibodies (MAbs) against mouse CFAP161 were generated as described in 33 by immunisation of rats with the peptide RMGNWNEDVYLEEERM (pI; aa11-26). MAbs that reacted specifically with CFAP161 were further analysed by Western blots and indirect immunofluorescence. α-CFAP161 pI clone 8F9 was further characterised (see Fig. S3 and "Supplemental Methods" for details) and used in this study. Rabbit polyclonal antibodies against mouse CFAP161 epitope ELLIQRNRRVKKNIL (pII; aa37-51) and YLD-SHKVEKPKNQW (pIII; aa254-267) in two different animals (a1 and a2) were generated and affinity purified by BioGenes (BioGenes GmbH, Berlin).
Yeast two-hybrid (Y2H) assay to validate CFAP161-KIAA0556 interaction. Yeast two-hybrid assays were performed according to the manufacturer's protocol (Matchmaker Gold Yeast Two-Hybrid System 630489, Clontech). For details see "Supplemental Methods".
Immunoprecipitations (IPs) and mass spectrometry (MS). CFAP161 was immunoprecipitated from wild type mouse testis with epididymis using the Thermo Scientific Pierce crosslink IP kit using α-CFAP161 pII or with rabbit-IgG as controls. In each experiment (n = 3) the total lysate of one homogenised (160-200 mg) testis and epididymis (~ 8-10 mg total protein) was incubated either with IgG or affinity-purified α-CFAP161 pII crosslinked to resin. 1 mg of protein was loaded to 40 µl settled resin as suggested by the manufacturer's instructions (in practice, 2 mg of total protein was loaded per column containing 80 µl of resin) in n = 6 IgG control IPs and n = 7 CFAP161 IPs. Immunoprecipitated CFAP161 complexes were isolated, purified, and subsequently analysed by LC-MS/MS as described 34  CFAP161/Cfap161 localisation. 20 pg of either pCS2 + EmGFP-Cfap161, pCS2 + HA-cfap161 or pCS2 + cfap161-HA were co-injected with 20 pg pCS2 + cetn4-RFP or pCS2 + GFP-cetn4 at the 4 cell stage into the ventral marginal zone to target the epidermis. Embryos were fixed in MEMFA and stained with Thermo Fisher Scientific Alexa Fluor 405 Phalloidin to mark the F-actin of the cell borders before fluorescence imaging. CFAP161-and CETN4-localisation was shown by direct fluorescence.
High-speed video microscopy of larval epidermal cilia. Videos were recorded at stage 32 embryos using a Zeiss Axioskop 2 mot plus microscope. Embryos were mounted on a slide containing a chamber constructed from duct tape. The most ventral part of the belly was recorded with a high-speed Hamamatsu video camera Orca flash 4.0 at 800 frames per second (fps) for 1 s to analyse ciliary beating. Ciliary flow was analysed using 1 μm fluorescent beads (Invitrogen FluoSpheres; 1:2000). Beads were added to the culture medium (0.1 × MBSH) and specimens were imaged using a Zeiss Axiocam HSm camera at 175 fps. Evaluation of CBF and bead transport was previously reported in 33 .  56 . All trajectories were further analysed using a custom-made program written in RStudio which calculated the velocity of each particle track 57 .

Data availability
The data generated or analysed during this study are included in this article (and its Supplementary Information files). The full mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 58 partner repository with the dataset identifier PXD022420. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.