Arginine‐ but not alanine‐rich carboxy‐termini trigger nuclear translocation of mutant keratin 10 in ichthyosis with confetti

Abstract Ichthyosis with confetti (IWC) is a genodermatosis associated with dominant‐negative variants in keratin 10 (KRT10) or keratin 1 (KRT1). These frameshift variants result in extended aberrant proteins, localized to the nucleus rather than the cytoplasm. This mislocalization is thought to occur as a result of the altered carboxy (C)‐terminus, from poly‐glycine to either a poly‐arginine or ‐alanine tail. Previous studies on the type of C‐terminus and subcellular localization of the respective mutant protein are divergent. In order to fully elucidate the pathomechanism of IWC, a greater understanding is critical. This study aimed to establish the consequences for localization and intermediate filament formation of altered keratin 10 (K10) C‐termini. To achieve this, plasmids expressing distinct KRT10 variants were generated. Sequences encoded all possible reading frames of the K10 C‐terminus as well as a nonsense variant. A keratinocyte line was transfected with these plasmids. Additionally, gene editing was utilized to introduce frameshift variants in exon 6 and exon 7 at the endogenous KRT10 locus. Cellular localization of aberrant K10 was observed via immunofluorescence using various antibodies. In each setting, immunofluorescence analysis demonstrated aberrant nuclear localization of K10 featuring an arginine‐rich C‐terminus. However, this was not observed with K10 featuring an alanine‐rich C‐terminus. Instead, the protein displayed cytoplasmic localization, consistent with wild‐type and truncated forms of K10. This study demonstrates that, of the various 3′ frameshift variants of KRT10, exclusively arginine‐rich C‐termini lead to nuclear localization of K10.

Keratins comprise the components of the epithelial cytoskeleton. Acidic (type I) and basic (type II) keratins heterodimerize and subsequently assemble to form intermediate filaments (IFs). 6 Epidermal keratinocytes specifically co-express keratins as pairs in a differentiation-dependent manner. Keratinocytes of the proliferative basal layer express keratin (K) 5 and K14. The differentiating suprabasal epidermal layers are associated with the down-regulation of K5/K14 and expression of thicker K1/K10 filament bundles. 7 All currently described IWC-associated variants are deletions, insertions or duplications located within the helix-termination-motif or in the V2 domain of either K10 or K1 (Table 1). Each results in frameshifts, leading to the translation of aberrant-truncated protein tails. Nonsense variants in the V2 domain of either K10 or K1 have not been reported. Variants leading to −2/+1 frameshifts within regions encoding the V2 domain are predicted to alter the downstream amino acid code from native glycine-to arginine-rich sequences.
Arginine-rich sequences frequently function as nuclear localization signals. 8,9 Therefore, aberrant arginine-rich K10 or K1 tails might be responsible for the pathogenic nuclear localization of K10 or K1, characteristic of IWC.
Two studies suggested that alanine-rich K10 tails might also lead to nuclear K10 localization. 10,11 In contrast to arginine, alanine-rich sequences have not been described to function as nuclear localization signal. 12 Lim et al, 10 suggested that a KRT10 frameshift variant at the exon 6/intron 6 boundary resulted in an alanine-rich tail. However, the original c.1373delG nomenclature for this variant should rather be annotated as c.1373+1delG, according to the sequence variant nomenclature HGVS, 13  resulted in an alanine-rich K10 tail. However, the variant was later on corrected to c.1544dupG and has been suggested to result in an arginine-rich K10 tail. 14 In support of a non-nuclear localizing signal of alanine-rich keratin tails, frameshift variants in KRT1, leading to alanine-rich tails, are reported to cause ichthyosis hystrix Curth-Macklin (IHCM) 15,16 and striate palmoplantar keratoderma (SPPK). 17 Keratinocytes of these patients do not display nuclear localization of aberrant K1 products. 16 The aim of this study was to clarify the effects of mutant K10 tail variants on the pathogenic nuclear localization of the protein.
In order to elucidate this, we established two distinct cell culture models. This study elucidates the relationship between K10 C-termini variants and K10 nuclear localization. The influence of C-termini variants on keratinocyte differentiation, K10 polymerization with assembly partners and the intracellular localization of these polymers were additionally characterized.

| Plasmids and transfection
cDNA-derived KRT10 was PCR-generated from epidermal mRNA isolated from an IWC patient 3 (p5_arg_c_GFP) or a healthy control (p2_ wt_c_GFP). This was cloned into XhoI and HindIII sites of the pEGFP-C1 vector (Clontech Laboratories Inc Takara). Sequences encoding the alanine-rich C-terminus (p10_ala_c_GFP) or a premature stop (p13_ter_c_ GFP) were inserted into p2_wt_c_GFP via site-directed mutagenesis (GenScript ® ). Genomic DNA-derived KRT10 from a healthy control was cloned into pUC19 in two steps following amplification of KRT10 as 3′ and 5′ products. Initially, intron 5 to 3′-UTR gDNA was inserted into the HincII and HindIII restriction sites of pUC19 (pUC_K10-ex5-3′).
Subsequently, the 5′ product (mid of exon 1 to exon 4) was introduced into the SacI and HincII restriction sites of pUC_K10-ex5-3′. This combined product (pUC-K10) was inserted into the SacI and HindIII restriction sites of p2_wt_c_GFP, exchanging cDNA between mid of exon 1 to 3′-UTR for a complete gDNA sequence. This wild-type gDNA construct (p1_wt_g_GFP) contained a common 12 bp deletion in exon 7 (rs778613907) 5  using XhoI and HindIII restriction sites ( Figure 1A). To distinguish the KRT10 wt , KRT10 arg and KRT10 ala products following combined transfection in cell imaging analyses, eGFP in p6_arg_s_GFP and p11_ala_s_GFP was replaced with mCherry. Constructs are summarized in Table S1 and Figure S1. NKc21 keratinocytes were transfected with Xfect transfection reagent (Takara Bio Europe) as previously described. 18 Transfected cells were seeded on coverslips 48 hours post-treatment and subsequently fixed 24 hours later with 4% formaldehyde. Transfection rates were approximately 30%.
Mouse 3T3-J2 fibroblasts (ATCC) were grown in DMEM (Lonza) supplemented with antibiotic-antimycotic and 10% foetal calf serum (ThermoFisher Scientific Inc). Feeder cells were generated from this cell line following mitomycin C (StressMarq Biosciences) treatment (4 mg/mL for 2 hours). These were subsequently used in keratinocyte single-cell cloning. Corresponding sequence data on KRT10 are summarized in Table S2.

| Keratinocyte differentiation and epidermal equivalent formation
Since KRT10 is only expressed in differentiated cells, differentiation was induced in confluent cultures in the absence of growth factors and at high calcium concentration (1.2 mM). For mRNA isolation, keratinocytes were cultured in CnT-PR medium (CELLnTEC). One day before expected confluency, medium was exchanged to CnT-PR-D ANOVA tests were used for statistical analysis.

| Ratio of mRNA p9_arg_s_GFP and p11_ala_s_ GFP
AB 3130 Genetic Analyzer was used to determine the ratio of KRT10 arg (p9_arg_s_GFP) and KRT10 ala (p11_ala_s_GFP) mRNA within PCR products ( Figure 1) by fragment analysis as previously described. 22 In brief, PCR was performed on cDNA following p4_var_g_GFP transfection with one FAM-labelled primer. This was stopped after 30 cycles to stay in the log phase of amplification. GeneScan™ 500 ROX™ (ThermoFisher Scientific Inc) was used as size standard. Quantitative assessment of PCR product was achieved following calculation of areas under the curve using GeneMapper.

| Immunofluorescence staining
Immunofluorescence staining was performed as previously described both on keratinocytes containing the respective plasmid grown on coverslips and on formalin-fixed and paraffin-embedded (FFPE-slides) keratinocytes of the epidermal models. 3 In brief, samples were blocked in donkey serum and triton-X-100 in TBS for 1 hour and incubated with primary antibodies either at RT for 1 hour (coverslips) or at 4°C overnight (slides).

| IWC-associated KRT10 variants result in an arginine-rich C-terminus
NKc21 keratinocytes were transfected with KRT10 gDNA constructs, featuring either wild-type (p1_wt_g_GFP) or a previously described IWC-associated splice site variant (p4_var_g_GFP) 10 to investigate the presence and composition of potential additional splice products (Table S1). The wild-type construct resulted in the transcription of one main mRNA product of the expected size (2.1 kb) (wt_s). In contrast, the construct encoding the IWCcausing sequence resulted in four major mRNA products, of differing sizes ( Figure 1A). Each mRNA was distinct (Table S1, Figure   S1). p14_ter_s_GFP) contained the complete intron 6, resulting in a shifted reading frame with a stop codon after 4 bp of intron 6.
The fourth product contained two sequences, with the majority of products displaying a 5 bp deletion of the 3′-terminus of exon 6 (r.1369_1373del, p.Gly457PhefsTer118; subcloned in p9_ arg_s_GFP) leading to K10 arg expression. A minority of products displayed a deletion of the last nucleotide of exon 6 (r.1373del; subcloned in p11_ala_s_GFP), encoding K10 ala (p.Ser458IlefsTer157).
Co-transfection with combinations of distinct fluorescence-labelled KRT10 arg and KRT10 ala (p6_arg_s_GFP/p12_ala_s_mCherry or p7_arg_s_mCherry/p11_ala_s_GFP) resulted in an expected nuclear signal of K10 arg . However, the unexpected nuclear localization of K10 ala was also observed ( Figure S3). Co-transfection with distinct fluorescence-labelled K10 arg (p7_arg_s_mCherry) and K10 wt (p3_wt_s_GFP) resulted in nuclear-localized eGFP. This indicated co-translocation of K10 wt with K10 arg into the nucleus. Further, it may suggest the nuclear translocation of higher keratin polymers ( Figure S3).
Immunostaining of cellular K5 and K14 after NKc21 transfection with a K10 arg vector (p5_arg_c_GFP) revealed a nuclear signal for endogenous K5 in approximately 30% of all transfected cells ( Figure   S4). K5 is a type II keratin, a promiscuous assembly partner of the   (Table S2).

| Endogenous expression of aberrant K10 leads to abnormal differentiation of keratinocytes in culture
Epidermal equivalents derived from K10 arg single-cell clones were subjected to haematoxylin and eosin (HE) staining. These displayed stratified epithelium including: basal layer, differentiated suprabasal layers and stratum corneum comparable to wild-type and mock controls ( Figure 5A-C). However, epidermal equivalents derived from K10 ala clones displayed an epithelium with strongly impaired differentiation and significantly thinner than epidermal equivalents from K10 arg and K10 wt ( Figure 5D, Figure S5).
Immunofluorescence analysis of differentiated keratinocytes via either N-or C-terminus-specific anti-K10 antibodies indicated a nuclear localization of K10 signal in K10 arg keratinocytes. K10 aladifferentiated keratinocytes displayed cytoplasmic signals alone, comparable to K10 wt ( Figure 6). Immunostaining via the C-terminus-specific K10 antibody enabled exclusive staining of K10 wt .
This revealed nuclear localization of K10 wt in the differentiated suprabasal layers of epidermal equivalents derived from heterozygous K10 arg keratinocyte single-cell clones. Similar observations were noted following co-immunostaining for K5. This supports our conclusion that, minimally, keratin tetramers driven by K10 arg are subjected to nuclear co-localization ( Figure S6). Additionally, lamin B1 displayed reduced expression in K10 ala expressing keratinocytes. This may indicate a disturbed nuclear integrity in affected keratinocytes.
However, the effect of K10 ala was previously inconclusive.
These affect splice sites in 60% of cases. Of the currently reported IWC-associated KRT10 splice site variants, approximately 30% affect the donor splice site of intron 6. Analysis of mRNA from patients with IWC-associated splice site variants revealed transcripts that would be translated into K10 with arginine-rich tails. 1   F I G U R E 6 Epidermal equivalents derived from K10 arg keratinocyte single-cell clones display nuclear localization of K10 in differentiated suprabasal layers. Immunofluorescence staining via N-(LH2, exact binding region unknown; red) and C-terminus-specific (EP1607IHCY, amino acids 555-584; green) K10 antibodies indicated nuclear localization of K10 in the suprabasal keratinocytes of K10 arg -derived equivalents (white arrowhead). In K10 wt -or K10 ala -derived equivalents, K10 was localized exclusively in the cytoplasm. Nuclear cotranslocation of endogenous K10 wt and K10 arg (e4_arg_26-1) enabled observation of nuclear localization via the C-terminus-specific anti-K10 (white arrowhead). DAPI (blue) was used for nuclear staining. Border between supporting filter membrane and basal layer indicated (white-dotted line). Scale bar, 20 μm an alanine-rich C-terminus results in the severe phenotype of IHCM 15,16,29 or SPPK, 17 no patients have been described with K10 ala -associated variants. This is in line with our data of an impaired differentiation in K10 ala keratinocytes, which might result in lethality.
Keratins promiscuously form heterodimers consisting of equimolar amounts of type I and type II keratins. 6 These subsequently polymerize to form keratin IFs. K1 is present within the nucleus of K10 arg -positive keratinocytes of IWC patients. 1 This is likely co-translocated with K10 arg in the form of dimers. In line with this observation, we observed nuclear co-localization of K5 with K10 arg in our model systems. This has also been observed in patient keratinocytes (unpublished data). Further, we observed nuclear co-localization of K10 arg /K10 wt in addition to K10 arg / K10 ala . This may be a result of K10 arg -containing tetramers or higher-order polymers passing the nuclear membrane. However, K14 was unexpectedly not detectable in the nuclear keratin complexes, possibly a result of poor epitope recognition by the monoclonal antibody.
In summary, this study provides a deeper understanding of the nuclear localization of aberrant K10 in IWC which is correlated to a single altered reading frame in K10. We confirmed that argininerich K10 tails, and not alanine-rich K10 tails, are responsible for the pathogenic nuclear localization of K10 in affected patients. A greater understanding of keratin nuclear localization in IWC advances the potential use of the subsequent mechanism of chromosomal exchange for natural gene therapies.

ACK N OWLED G EM ENTS
We thank Hans Törmä (Uppsala, Sweden) for providing the NKc21 cell line and Irene Leigh and Andrew South for providing the antibody LH2. We are grateful to our Pathology Department for embedding the epidermal equivalents and Nicholas Sanderson (Basel, Switzerland) for support in cloning. We also thank our microscope and FACS facilities for support. The study was ethically approved by the local ethics committee (Ethikkommission Nordwest-und Zentralschweiz) (EK203/13) and financially supported by Louis Widmer Fonds (Schlieren, Switzerland).

CO N FLI C T O F I NTE R E S T
Nothing to disclose.

AUTH O R CO NTR I B UTI O N S
PR, EI, IS, HW and SvA performed the experiments. MA, OPM and AV contributed essential tools. PR, AV and BB analysed the data. BB, JR and OPM wrote the paper. PHI, JR and BB designed the study.

DATA AVA I L A B I L I T Y S TAT E M E N T
I am willing to share my research data.