Inactivation of DNase1L2 and DNase2 in keratinocytes suppresses DNA degradation during epidermal cornification and results in constitutive parakeratosis

The stratum corneum of the epidermis constitutes the mammalian skin barrier to the environment. It is formed by cornification of keratinocytes, a process which involves the removal of nuclear DNA. Here, we investigated the mechanism of cornification-associated DNA degradation by generating mouse models deficient of candidate DNA-degrading enzymes and characterizing their epidermal phenotypes. In contrast to Dnase1l2 −/− mice and keratinocyte-specific DNase2 knockout mice (Dnase2 Δep), Dnase1l2 −/− Dnase2 Δep mice aberrantly retained nuclear DNA in the stratum corneum, a phenomenon commonly referred to as parakeratosis. The DNA within DNase1L2/DNase2-deficient corneocytes was partially degraded in a DNase1-independent manner. Isolation of corneocytes, i.e. the cornified cell components of the stratum corneum, and labelling of DNA demonstrated that corneocytes of Dnase1l2 −/− Dnase2 Δep mice contained DNA in a nucleus-shaped compartment that also contained nucleosomal histones but lacked the nuclear intermediate filament protein lamin A/C. Parakeratosis was not associated with altered corneocyte resistance to mechanical stress, changes in transepidermal water loss, or inflammatory infiltrates in Dnase1l2 −/− Dnase2 Δep mice. The results of this study suggest that cornification of epidermal keratinocytes depends on the cooperation of DNase1L2 and DNase2 and indicate that parakeratosis per se does not suffice to cause skin pathologies.

cuticle of the inner root sheeth of hair follicles 15 . However, the stratum corneum of DNase1L2 knockout mice is orthokeratotic, suggesting that DNase1L2 is not essential for DNA degradation during cornification of interfollicular keratinocytes in mice 14 . In contrast to the results of DNase1L2 knockdown, the suppression of the lysosomal DNase2 abrogated acid DNase activity in human skin equivalents but did not lead to parakeratosis 16 . Mice in which Dnase2 was inactivated by the Cre-loxP system in keratin K14-positive cells and their progeny cells (Dnase2 Δep ), including epidermal keratinocytes and sebocytes, displayed incomplete DNA degradation during holocrine secretion of sebum but did not result in parakeratosis in the interfollicular epidermis 15 .
Here, we investigated whether DNase1L2 and DNase2 act together in mouse stratum corneum formation. We show that Dnase1l2 −/− Dnase2 ∆ep mice develop parakeratosis and suggest that DNase1L2, DNase2, and at least one more DNA-degrading enzyme degrade DNA during cornification of keratinocytes.

Material and Methods
Mice. The generation of Dnase1l2 −/−14 , Dnase1 −/−17,18 , Dnase2 ∆ep , Dnase1l2 −/− Dnase2 ∆ep , and Dnase1 −/− Dnase1l2 −/− Dnase2 ∆ep15 mice was reported previously. Dnase1 knockout and Dnase1l2 knockout mice were derived from breeding colonies described previously 14,17 . In Dnase2 ∆ep , also referred to as Dnase2 f/f K14-Cre mice, essential exons of the Dnase2a gene are flanked by loxP sites (Dnase2 f/f ) 19   Preparation and investigation of corneocytes. Corneocytes were prepared by tape stripping of the skin surface using D-squames ® (Cuderm, Dallas, TX). The corneocytes attached to the tape were subjected to DNA-labeling and visualized by fluorescence microscopy and bright-field microscopy with phase contrast. The corneocyte stress test was performed as described previously 22 with modifications. In brief, the epidermis of ear skin was separated from the dermis by treatment with dispase and subsequently incubated in PBS containing 2% SDS at a temperature of 95 °C for 10 min. The remaining corneocytes were centrifuged and resuspended in the same buffer at a density of 10 6 /ml. The corneocyte suspensions were sonicated in a sonicator (Hielscher UIS250L, Teltow, Germany). Aliquots were removed every 10 seconds, and intact corneocytes were counted with a counting chamber under the microscope. At least 3 biological replicates were measured. Differences were evaluated with the two-sided t-test. P-values greater than 0.05 were considered not significant.
Data availability statement. The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.

Results
Co-deletion of DNase1L2 and DNase2 causes epidermal parakeratosis. H&E staining and DNA labelling with Hoechst 33258 dye showed that, in agreement with previous reports 14,15 , wild-type mice as well as Dnase1l2 −/− , and Dnase2 f/f K14-Cre mice (here termed Dnase2 ∆ep mice with ∆ep short for deletion in epithelial cells derived from K14-positive precursors) had an orthokeratotic stratum corneum (Fig. 1A,C,E,G, and data not shown). Mice lacking both DNase1L2 and DNase2 in epidermal keratinocytes, were generated by crossing Dnase1l2 −/− mice with Dnase2 Δep mice. H&E staining showed distinct nuclear remnants in the stratum corneum (parakeratosis) on the soles and footpads of Dnase1l2 −/− Dnase2 Δep mice (Fig. 1B). No nuclear remnants were detected within the stratum corneum on the ear and back skin after H&E staining (Suppl. Fig. S1). However, DNA labelling with Hoechst 33258 dye revealed the presence of nuclear remnants within the stratum corneum also of these body sites of Dnase1l2 −/− Dnase2 Δep mice (Fig. 1D,F,H). In line with the H&E stainings, the aberrant retention of DNA was most prominent in the epidermis of the footpads (Fig. 1D).
Scientific RepoRTs | 7:6433 | DOI:10.1038/s41598-017-06652-8 DNA in parakeratotic stratum corneum of Dnase1l2 −/− Dnase2 Δep mice is fragmented. The TUNEL assay showed that parakeratotic stratum corneum of Dnase1l2 −/− Dnase2 Δep mice contained free 3′-OH ends, indicating that the retained DNA was at least partially fragmented (Fig. 2). As DNase1 is a candidate enzyme for generating these DNA ends, we generated DNase1 −/− Dnase1l2 −/− Dnase2 Δep mice. Deletion of DNase1 alone did not result in parakeratosis, and deletion of DNase1 in addition to the deletion of DNase1L2 and DNase2 did not abrogate the TUNEL-positivity of parakeratotic corneocytes (Suppl. Fig. S2 and data not shown). These results argue against a critical role of DNase1 in cornification. Thus, the partial degradation of DNA in the stratum corneum of Dnase1l2 −/− Dnase2 Δep mice is due to an as yet unknown mechanism. Immunofluorescence analysis of wild-type mice demonstrated that normal cornification involves the complete removal of histones H1 and H3 (Fig. 3). By contrast, in DNase2/DNase1L2-deficient keratinocytes histone H3, which is part of nucleosomes, was retained during cornification (Fig. 3C,D) while the internucleosomal histone H1 (Fig. 3A,B) and the nuclear intermediate filament protein, lamin A/C (Fig. 3E,F) behaved like in wild-type mice. These data suggest that DNA degradation is a necessary prerequisite for enabling degradation of histone H3 during normal cornification whereas nuclear proteins that are not enwrapped by DNA are degraded independently of DNase activities.
DNase1L2/DNase2-deficient corneocytes retain DNA in a nucleus-shaped compartment, display normal resistance to stress and establish a functional skin barrier. Next, we isolated corneocytes from Dnase1l2 −/− Dnase2 Δep and wild-type mice, and incubated them with DNA-specific dye. Corneocytes from Dnase1l2 −/− Dnase2 Δep mice consistently contained DNA that appeared to be concentrated in nucleus-shaped compartments (Fig. 4). Bright field microscopy revealed that corneocytes isolated from Dnase1l2 −/− Dnase2 Δep mice remained intact during the isolation procedure and had morphologies comparable to those from wild-type mice (Fig. 4). Exposure to ultrasound also resulted in comparable decay of wild-type and DNase1L2/DNase2 double knockout corneocytes (Suppl. Fig. S3), indicating that the retention of DNA did not increase the sensitivity of corneocytes to mechanical stress.
Measurement of the trans-epidermal water loss (TEWL) in wild-type and Dnase1l2 −/− Dnase2 Δep mice did not reveal significant differences between the two genotypes (Suppl. Fig. S4). In situ immunostaining for inflammatory cells in the skin (Suppl. Fig. S5) showed no significant difference between Dnase1l2 −/− Dnase2 Δep and wild-type mice (Suppl. Fig. S5). These results suggest that parakeratosis due to DNase2/DNase1L2-deficiency of keratinocytes does neither impair the barrier function of the stratum corneum nor trigger an inflammatory reaction in the skin.

Discussion
In this study both DNase1L2 and DNase2 were identified as enzymes that participate in the degradation of DNA during cornification in the interfollicular epidermis of mice. Previously, we have generated DNase1L2 knockout mice which display aberrant retention of DNA in the corneocytes of hair fibers, inner root sheaths of hair follicles, nails, tail scales, and filiform papillae of the tongue 14,15 . We have also generated Dnase2 Δep mice which show aberrant retention of DNA in sebocytes and in the isthmus of the hair follicle 12 . But none of these two mouse strains had parakeratosis in the interfollicular epidermis of the back and ear skin nor on the plantar skin. The absence of epidermal parakeratosis in DNase1L2 knockout mice was surprising because suppression of DNase1L2 alone sufficed to cause parakeratosis in an in vitro model of human skin 12 . The results of the present study suggest that DNase1L2 contributes to DNA degradation also during stratum corneum formation in the mouse. However, only when both DNase1L2 and DNase2 are inactivated, DNA degradation in corneocytes remains incomplete. Thus, murine cornification-associated DNA degradation appears to be a multi-step process in which the abrogation of either DNase1L2 or DNase2 activities alone, but not the suppression of both enzymes, can be compensated (Fig. 5).
Our finding that DNA in the parakeratotic stratum corneum of Dnase1l2 −/− Dnase2 Δep mice is TUNEL-positive suggests that it is partly degraded. We could exclude that this degradation depends on DNase1, which is also expressed in differentiating keratinocytes 23 . Other candidate contributors to cornification-associated DNA fragmentation include, but are not limited to, caspase-activated DNase (CAD), also known as DNA fragmentation factor B 24 and the exonuclease TREX2 10,25,26 .
Like in suppression of DNA breakdown during programmed cell death of hair and nail keratinocytes 14 and sebocytes 15 , the blockade of DNA degradation by deletion of DNase1L2 and DNase2 prevented the complete removal of histone H3, a component of nucleosomes, whereas the inter-nucleosomal histone H1 and a nuclear protein without DNA interaction, lamin A/C, were degraded. Of note, the immunoreactivity of H3 was reduced in the lower but not in the upper layers of the stratum corneum in Dnase1l2 −/− Dnase2 Δep mice (Fig. 3D), suggesting that the epitope targeted by the anti-H3 antibody was transiently masked by a cornification-related mechanism. We propose that DNA degradation via differential contributions of DNase1L2 and DNase2 exposes histone H3, which is otherwise physically protected by DNA, to the action of proteases in terminally differentiating keratinocytes that give rise to three distinct products, i.e. hard skin appendages 14 , sebum 15 , and the stratum corneum (this study). Interestingly, a previous report has suggested that AKT1 signaling and degradation of lamin A/C are essential in the cascade of events that initiate DNA degradation during cornification 27 . Our data suggest that the breakdown of lamin is not sufficient to allow for the breakdown of the nucleus.
The persistence of DNA in keratinocytes undergoing programmed cell death was not associated with an inflammatory reaction in the skin of Dnase1l2 −/− Dnase2 Δep mice. This suggests that the aberrantly retained DNA did not reach or activate sensors of mislocalized DNA, such as TLR9, AIM2, and cGAS [28][29][30] . The DNA appeared to be concentrated in the center of corneocytes of Dnase1l2 −/− Dnase2 Δep mice (Fig. 4) whereas the periphery of these corneocytes appeared to devoid of DNA, perhaps indicating a physical or chemical mechanism for DNA retention in the position of the nucleus or a DNase1L2 and DNase2-independent mechanism for DNA-degradation in the cytosolic part of corneocytes. The cornified envelope represents another likely barrier against the leakage of DNA out of parakeratotic corneocytes which is a prerequisite for reaching receptors for DNA in living cells of the skin. Alternatively, it is conceivable that DNA in cornifying DNase1L2/DNase2 double knockout keratinocytes is partly taken up by other cells without activating inflammation.
Parakeratosis is an important phenomenon in the diagnosis of many skin diseases 3 , but it has remained unclear whether the retention of nuclear constituents per se may contribute to the pathogenesis of skin diseases [31][32][33] . The results of this study demonstrate that the presence of nuclear DNA in association with histones in corneocytes does not suffice to cause pathological changes. It remains to be investigated whether aberrantly retained DNA in cornifying mouse keratinocytes has a pro-inflammatory effect when the epidermis is exposed to different types of stress. Finally, it is possible that the mouse model does not replicate all relevant processes active  in human skin. The expression of DNase1L2 is downregulated in parakeratotic skin lesions of psoriasis patients 12 and DNA degradation activity at neutral pH, corresponding to DNase1 family members, is reduced in psoriatic scales 34 . By contrast, DNA-degradation activity at acidic pH, corresponding to DNase2, is present in extracts from psoriatic scales 34 , indicating that DNase2 is expressed in sufficent amounts in psoriatic skin. However, it is currently unknown whether DNase2 is proteolytically activated in aberrantly differentiated epidermis 35 or whether the elevated pH in psoriatic skin, as compared to normal skin 36 , reduces DNase2 activity. Further studies in human skin are necessary to determine whether suppression of DNase1L2 and DNase2 activities contribute to parakeratosis in human skin diseases and to evaluate the response to incomplete DNA degradation in cornifying human keratinocytes.
Our results establish the concept that multiple DNases, including DNase1L2 and DNase2, participate in cornification-associated degradation of nuclear DNA in the interfollicular epidermis, and provide a mechanistic basis for studying the roles of parakeratosis in clinical studies.