P-Selectin preserves immune tolerance in mice and is reduced in human cutaneous lupus

Mice deficient in P-Selectin presented altered immunity/tolerance balance. We have observed that the absence of P-Selectin promotes splenomegaly with reduced naïve T cell population, elevated activated/effector T cell subset, increased germinal center B and Tfh populations and high production of autoreactive antibodies. Moreover, 1.5-3-month-old P-selectin KO mice showed reduced IL-10-producing leukocytes in blood and a slightly reduced Treg population in the skin. With aging and, coinciding with disease severity, there is an increase in the IL17+ circulating and dermal T cell subpopulations and reduction of dermal Treg. As a consequence, P-Selectin deficient mice developed a progressive autoimmune syndrome showing skin alterations characteristic of lupus prone mice and elevated circulating autoantibodies, including anti-dsDNA. Similar to human SLE, disease pathogenesis was characterized by deposition of immune complexes in the dermoepidermal junction and renal glomeruli, and a complex pattern of autoantibodies. More important, skin biopsies of cutaneous lupus erythematosus patients did not show increased expression of P-Selectin, as described for other inflammatory diseases, and the number of vessels expressing P-Selectin was reduced.

Systemic lupus erythematosus (SLE) is a chronic, inflammatory autoimmune disease characterized by the production of autoantibodies against double strand DNA (dsDNA) and nuclear antigens, immune complex deposition, complement activation and polyclonal expansion of autorreactive lymphocytes 11,12 . SLE predominantly affects women (6-10:1 ratio of women to men) in the childbearing years 12,13 . Clinical manifestations of SLE include inflammation of the skin and internal organs, which are translated into non-specific symptoms like fever, arthralgia, skin rashes and anemia 12 . P-Selectin levels are elevated in the urine of SLE patients and correlate with disease severity 14 . Genome-wide linkage studies in humans have suggested an important role for P-Selectin in SLE. Indeed, the P-Selectin gene is located in the SLE linkage region on human chromosome 1 (1q23) 15,16 . Moreover, variations in the upstream region of P-Selectin are a risk factor for SLE, and two risk alleles have been identified potentially affecting the transcription of P-Selectin and the binding to P-Selectin glycoprotein ligand-1 (PSGL-1) 15 , the main ligand for P-Selectin expressed on all leukocyte subsets, and also a ligand for Eand L-Selectin 3, [17][18][19] .

Results
Presence of circulating autoantibodies and augmented splenic reactivity in P-Sel −/− mice. Sera from P-Sel −/− mice contained anti-cytoplasmic and anti-nuclear autoantibodies with a speckled or mitotic pattern (Fig. 1a), but not anti-centromere autoantibodies. ELISA identified several antigens recognized by these autoantibodies, including topoisomerase I (Scl-70), U1-RNP, Sm and t-RNA synthetase (Jo-1) (Fig. 1b). Sera were also positive in the Crithidia assay for anti-dsDNA autoantibodies (Fig. 1c), a hallmark of human lupus erythematosus 1,3,5 , whereas none of the WT sera were positive in the same conditions. The anti-Sm, anti-topoisomerase I, and anti-dsDNA antibodies could be already detected at 3 months of age. The autoantibodies were co-expressed in the same animals and the percentage of mice with autoantibodies increased as a function of age, reaching a maximum at 12-18 months (Fig. 1b) and decreasing at 18-24 months, especially anti-Sm and anti-Scl-70 (Fig. 1b). The prevalence of anti-dsDNA autoantibodies increased gradually with ageing, from 10% in the 3-month-old group to 30% in the 18-24-month-old group (Fig. 1c). Consistent with this exacerbation of humoral immunity, we observed a remarkable spleen enlargement in both male and female 1.5-3-month-old P-Sel −/− mice (Fig. 1d), which has been described previously in lupus-prone mouse strains 30,31 . Considering the increased cellularity of 3 month-old P-Sel −/− spleens (Fig. 1d, lower panel), we analyzed the germinal centre (GC) B cell and the follicular helper T cell subsets (Tfh). We found an increase in both lymphocytic populations, indicating a more reactive state of P-Sel −/− spleens (Fig. 1e,f). Immune homeostasis imbalance in P-Sel −/− mice blood and spleen. The study of the peripheral blood leukocyte populations of 1.5-3-month-old male mice showed an important increase in the B cell compartment of P-Sel −/− mice. In contrast, the T cell compartment was reduced, mainly due to the decrease of the CD4 + T cell subset. Additionally, the monocytic population was also reduced in P-Sel −/− animals (Fig. 2a).
We studied the cytokine production of the circulating immune populations and we found a reduction in the percentage of IL-10 + producing cells of 1.5-3-month-old P-Sel −/− mice, that was statistically significant among cDC and CD4 + T lymphocytes and close to the statistical signification in the B cell population (Fig. 2b,c, left panels). Circulating leukocytes of > 18-month-old mice showed a statistically significant reduction in the percentage of IL-10 + producing monocytes and B cells (Fig. 2d, left and middle panels). Interestingly, the IL-17 + CD4 + (Th17) population was increased in aged P-Sel −/− mice (Fig. 2d, right panels).
Immune homeostasis imbalance in the skin of P-Sel −/− mice. Regarding the skin, one of the main organs affected in human lupus, we did not find significant changes in the different immune subpopulations in 1.5-3-month-old mice, except from an increased percentage of plasmacytoid DC (pDC) in P-Sel −/− animals ( Fig. 3a, upper panels). However, in aged animals, we found a reduced population of macrophages and a significant increment in the T cell and pDC subsets (Fig. 3a, lower panels). Among T cells, we did not find any difference in the percentage of gamma/delta (γ δ ) T cells (Fig. 3a, right panels). We studied the cytokine production of the different populations, and only the > 18-month-old P-Sel −/− mice showed a reduction in the percentages of IL-10 + macrophages, cDC, pDC and B cells (Fig. 3b, lower panels; Fig. 3c). Accordingly, aged knocked-out mice displayed a reduced FOXP3 + (Fig. 3d and e) and an increased Il-17 + T cell populations ( Fig. 3f and g). We did not found significant differences in cytokine production in 1.5-3-month-old mice.
The exposure of 3-4-month-old female WT and P-Sel −/− to UV radiation provoked an extensive dermatitis in P-Sel −/− mice with ulcers in the exposed skin, but not in WT counterparts (Fig. 4d). Histopathological analysis revealed a severe epidermal lesion in female P-Sel −/− mice consisting of acanthosis and hyperkeratosis, immune infiltration in the dermis, and deposition of extracellular matrix components in the hypodermal layer (Fig. 4e). Consequently, the severity of the lesions was significantly higher in UV-irradiated P-Sel −/− mice (P-Sel −/− 7.25 ± 2.99 vs WT 3.00 ± 1.41) (Fig. 4f).

Kidney alterations and deposits of immune complexes in skin and kidneys of P-Sel
Histological examination of kidney sections of WT and P-Sel −/− mice showed that P-Sel −/− mice had a high proportion of glomeruli with a dilated Bowman's space (Fig. 5a). Additionally, some glomeruli presented tubularization of the Bowman's capsule (Fig. 5a). We found also interstitial infiltrates in P-Sel −/− mice (Fig. 5b), whose prevalence increased from 40% at 1.5-3 months of age to 84% in mice older than one year (12-24 months-old) (Fig. 5c), while only 20% of aged WT and none of the young WT mice presented interstitial infiltration.
We also observed that 30% of the 3-month-old P-Sel −/− mice had infarcted foci (Fig. 5d). The prevalence of infarcts increased as mice grew older (Fig. 5e), reaching to 60% of females and 80% of males in the >18-month-old population of P-Sel −/− mice.
We found deposits of immune complexes in the glomerular basal membrane in 100% of female and 25% of male P-Sel −/− mice older than 18 months, but only in 25% of WT females and none in WT males (Fig. 5f). In the skin, we found deposits of immune complexes in 60% of male and 40% of female P-Sel −/− mice over 18 months of age but in none of WT mice (Fig. 5f).
According to the structural renal damage and the immune complex deposition, we found that 13% of > 18 months P-Sel −/− mice developed proteinuria, while 38% of P-Sel −/− mice develop hematuria (Fig. 5g).  Reduced lifespan of P-Sel −/− mice. To analyze the impact of the autoimmune syndrome progression during the lifespan of P-Sel −/− mice, we carried out a survival study with 20 WT and 23 P-Sel −/− mice from 6 to 100 weeks of age. P-Sel −/− mice started to die at week 47 and we observed a peak of mortality at around week 80. At week 100, 90% of WT mice remained alive whereas only 61% of the P-Sel −/− mice were still alive (Fig. 5h).

Decreased expression of P-Selectin in SLE skin biopsies.
To assess the relevance of P-Selectin in human lupus, we compared by immunohistochemical staining the expression of P-Selectin in endothelial cells of the dermal vessels of skin biopsies obtained from cutaneous lupus erythematosus (cLE) patients and healthy controls. We identified all the blood vessels (CD31 + ) in the whole biopsy and classified them into three categories depending on the expression of P-Selectin ( Fig. 6a and b): (1) unstained, (2) partially stained, and 3) fully stained. We found a deep reduction of fully stained blood vessels in cLE biopsies, which were accompanied by a remarkable elevation in the percentage of negative vessels for P-Selectin expression (Fig. 6c). We also found unspecific binding to infiltrating leukocytes, that has been already reported by other authors 36,37 .

Discussion
In this work, we show that the absence of P-Selectin breaks the immune tolerance and triggers the development of a progressive autoimmune lupus-like syndrome displaying most of the features previously described in lupus-prone mice. Importantly, we show that human biopsies of cutaneous lupus have reduced number of P-Selectin stained vessels.
Given that PSGL-1 deficient mice develop a scleroderma-like syndrome, we have analyzed whether the absence of P-Selectin, main ligand of PSGL-1, also triggered autoimmunity. We found the production of anti-dsDNA and anti-Sm autoandibodies, hallmarks of SLE in humans 38 . Interestingly, anti-Sm and anti-dsDNA autoantibodies are detected at an early age in P-Sel −/− mice. The high rate of autoantibodies probably favours, as in human SLE, the formation and deposition of immune complexes in anatomic sites characterized by high blood pressure including capillaries of the skin and glomeruli, among others. Accordingly, we found a high rate of immune complexes deposition in renal glomeruli and skin of aged P-Sel −/− which is a characteristic of both human and murine lupus 29,32,39,40 .
To understand the autoantibody production, we studied the impact of P-Selectin absence in the homeostasis of the immune system and analyzed the effector and regulatory leukocyte subsets in blood, spleen and skin of WT and KO mice, and found that the immune homeostasis is altered in P-Sel −/− mice. We found higher percentage of germinal center B cell and Tfh subpopulations, indicating more reactive germinal centers in the spleen of young animals as well as reduced circulating TCD4 + and monocytic populations, in association with an increment in the B cell compartment. Interestingly, at the young age we found a reduction in the circulating tolerogenic IL-10-producing cDC and monocyte subsets, as well as in the IL-10 + B cell compartment (Breg), whose reduction has been recently implicated in the development of human SLE 41 . Although young P-Sel −/− mice did not have altered the presence of Th1, Th2 and Th17 populations, reduction in the IL-10-producing CD4 + T cells implies an imbalanced Teff/Treg ratio, as it has been described for both murine and human SLE 42 . This reduction in IL-10 + cells could be explained by the lack of tolerogenic signal supplied by P-Selectin interaction with PSGL-1, as previously described 20 . Remarkably, we found a shift through a Th17 response with aging in the blood of P-Sel −/− mice coinciding with the worsening of the disease, in agreement with the described role of PSGL-1 expression on Treg in the attenuation of persistent T cell activation during the immune response 43 . In fact, IL-17 promotes the recruitment of T cells, monocytes and granulocytes to the inflammatory foci, activates B cells and contributes to the synthesis of IgG and anti-dsDNA autoantibodies 44 . Accordingly, we report that, in P-Sel −/− mice, the effector subpopulations of splenic CD4 + and CD8 + T cells were increased, with the consequent and relevant reduction among the naïve subset. SLE patients show reduced numbers of CD45RA + naïve T cells and increased numbers of CD45RO + memory T cells 45 . Interestingly, it has been described that mouse CD8 + CD122 + Treg have a central memory phenotype (CD44 high CD62L high ) 46 what could explain the reduction of this compartment in the P-Sel −/− mice, as well as the reduced populations of IL-10 producing T cells. In addition, this subset has been described to regulate T cell homeostasis and to suppress both autoimmune and alloimmune responses 46 .
Given that the skin is one of the main organs affected in lupus patients, we have also analyzed whether the dermal immune system is also altered in P-Sel −/− mice. Our results indicate that dermal pDC subpopulation is increased from youth and that aging also increases the T cell subset. Importantly, coinciding with the appearance of the lupus skin lesions, we have found a high reduction in the dermal IL-10 producing immune cells and the FOXP3 + Treg population, together with an important expansion of the IL-17 producing T lymphocytes. Our data are in accordance with the expanded Th17 population and reduced Treg subset described for human patients 47 . It has been described that different mechanisms can be implicated in the breakage of the immunological tolerance 48 and our work indicates that a defect in P-Selectin expression and function, probably through PSGL-1 interaction, could be implicated in maintaining the immune system immunity/tolerance balance. Dysfunction or low expression of P-Selectin and PSGL-1 could favor the use of alternative molecules for leukocyte extravasation and consequently, the loss of the tolerogenic signal triggered by PSGL-1/P-Selectin interaction.
The skin histological features described in this work for aged P-Sel −/− mice, such as hyperkeratosis, acanthosis and immune cell infiltration, have been previously reported for MRL/lpr mice 33,49 . According with our data, it has been reported that P-Selectin or PSGL-1 deficiency in MRL/lpr mice results in an increase in the severity of dermatitis and glomerulonephritis 29 . We observed that UV irradiation could accelerate and intensify the appearance of the above mentioned lesions in 3-4-month-old P-Sel −/− mice, whereas non lupus-prone strains are considered to be resistant to UV-induced DNA damage 38 . Interestingly, more than 50% of SLE patients show photosensitivity 38,50,51 .
Glomerulonephritis and renal involvement are the most common manifestations in SLE patients (40-70%) 52 . Leukocyte infiltration is a common feature shared by humans and lupus-prone mice 11 . Accordingly, we found leukocytic infiltration in both glomerular and tubulointerstitial compartments of the kidney in P-Sel −/− mice. Additionally, P-Sel −/− mice presented a high frequency of renal infarcts, indicating vascular dysfunction. Renal and systemic vasculitis have been described in patients and experimental models of SLE 27 and, together with immune infiltration, have been associated with loss of renal function 53 . In agreement with our data, mice deficient for P-Selectin are more sensitive to glomerulonephritis 29,54 . Importantly, as a consequence of the renal damage, we found increased prevalence of proteinuria and hematuria in P-Sel −/− mice. We also report that the lifespan of P-Sel −/− mice is reduced compared to WT animals, showing an augmented death rate at approximately 1 year of age (47 weeks), which is sharply increased at 80 weeks coinciding with disease worsening. This increased death along P-Sel −/− mice could underestimate the prevalence of some clinical observations such as anti-DNA autoantibodies, skin lesions or proteinuria.
Finally, it has been reported that the lack of either P-Selectin or PSGL-1 in murine lupus models enhances both skin and renal inflammation 29 and that the tissue expression of P-and E-Selectin in the MLR/lpr mice was not upregulated regarding the non-inflammed MRL +/+ mice 55 . Importantly, when analyzed in patients, although E-selectin expression was increased in skin biopsies 56 , according with our data in P-Sel −/− mice, we found a reduction in the expression of P-Selectin in the dermal blood vessels of patients with cutaneous lupus. This important result does not concur with the higher levels of P-Selectin that have been traditionally associated with other autoimmune and inflammatory diseases like glomerulonephritis, rheumatoid arthritis, psoriasis or atopic dermatitis 37,57,58 . It has been reported that TNFα and LPS downregulated human P-Selectin 7 . The increased levels of TNFα in the serum of SLE patients are consistent with the low expression of P-Selectin that we found in the skin of lupus patients.
In summary, we show that P-Selectin expression is crucial for the immune system homeostasis and that its absence promotes the spontaneous development of a lupus-like syndrome in mice. Accordingly, patients with cutaneous lupus showed lower expression of P-Selectin in the endothelium of the dermal vessels, suggesting that the reduced expression of P-Selectin could be implicated in the pathogenesis of this disease. As SLE pathogenesis is not well understood and there is not a universally curative treatment for lupus in humans, it is very important to discover new molecules implicated in the development of the different forms of this disease that could be used as targets for new treatments. Our work will contribute to the understanding of lupus pathogenesis, and our future research will explore the role of P-Selectin in the development of human lupus and other autoimmune diseases. In addition, our work suggests that P-Selectin KO mice could be used as a new experimental model for in vivo assays to evaluate new treatments or combination of treatments against the progression of the disease that could prevent organ damage associated with SLE. UV radiation protocol. Three month-old WT and P-Sel −/− mice were irradiated with an UVB lamp containing a set of six tubes (Phillips TL UV, 20 W; Royal Philips Electronics, Amsterdam, Netherlands) (ranging from 290 nm to 320 nm), being the energy output, at a distance of 15 cm, of 2.5 mW/cm 2 . Mice received three doses of 0.306 J/cm 2 every other day for a week and an additional dose of 0.12 J/cm 2 three days later. Skin was processed 24 hours after the last dose.

Mice
Autoantibody assays. The presence of anti-nuclear antibodies (ANAs), the Crithidia assay for anti-dsDNA and the characterization and quantification of circulating autoantibodies were performed as previously described 22 . The cut-off point for positivity on ELISA for a particular autoantibody was determined as the mean value (X) plus two standard deviations (SD) obtained from sera of at least 100 WT mice with an age ranging from1.5 to 24 months. At least 8 mice per group of age were analyzed.
Histopathological assessment and immunohistochemical (IHC) staining of mouse skin and kidney. Masson's trichrome staining was performed with the Artisan Gomori's Green Trichrome Stain kit (Dako; Glostrup, Denmark). Images were obtained with a Leica DM2500 light microscope and a Leica DFC450 camera. All sections were examined by a pathologist blinded to the sample origin.
Para-midline, upper back skin samples were evaluated blindly to obtain a semiquantitavive measurement by assigning a 3-criteria-based score 59 : acanthosis (0, normal; 1, slight thickening of epidermis; 2 and 3, presence of two or three layers of stratum spinosum cells, respectively; 4, presence of four or more layers of stratum spinosum cells); hyperkeratosis (0, normal; 1, one layer of keratin; 2, two layer thick of keratin; 3, minimum three-layer thick of keratin; 4, minimum three-layer thick of keratin and formation of a keratin's plug); hypergrannulosis (0, normal; 1, moderate accumulation of granular dark material in the stratum granulosum; 2 and 3, two or more than three layers of granular dark material; 4, three or more layers of accumulated granular dark material overall section of skin). Finally, the total pathology score was calculated by adding the resulting values for the three criteria.
The presence of kidney ischemic events (infarcted areas) was assessed in Masson's trichrome-stained sections. Skin and glomerular immunoglobulin deposition were evaluated by IHC with an antibody cocktail against Fab regions of IgA, IgG, IgM (1/250) (Abcam; Cambridge, UK), following manufacturer's instructions of the Universal LSAB+ Kit Rabbit/Mouse/Goat (DAB+ ) (Dako). At least 5 males and 5 females were evaluated per genotype and age.
Urine analyses. Urinary protein and haemoglobin were determined using Combur Test M dipsticks (Roche; Basel, Switzerland). The cut-off for positivity in the proteinuria test was the lowest value in P-Sel −/− mice not reached by any of the WT mice analyzed.
Scientific RepoRts | 7:41841 | DOI: 10.1038/srep41841 Human skin samples. Skin biopsies from 4 chronic/subacute cutaneous lupus erythematosus patients and from 4 aged-matched healthy controls were obtained from the Pathology Department, Hospital de la Princesa (Madrid, Spain). The investigations were conducted in accordance with the principles of the Declaration of Helsinki and were approved by the Clinical Investigation Ethical Committee of the Hospital de la Princesa, Madrid, Spain (Register number: PI-654, date of approval 07-02-2013). Informed consent was obtained from all the patients and healthy controls.
Immunohistochemistry of human skin sections. Consecutive tissue sections of sun-exposed skin biopsies underwent immunohistochemistry with the Dako REAL EnVision Detection System Peroxidase/DAB+ kit (Dako), using as primary antibodies anti-human P-Selectin (10 μ g/ml) (R&D Systems; Minneapolis, MN, USA) and anti-human CD31 (1/50) (Abcam). The whole biopsies were scouted for blood vessels in a blinded manner. Vessels were identified by positive staining by CD31, localized in the consecutive P-Selectin-stained section, and classified depending on the expression of P-Selectin as follows: 1) unstained, negative vessels; 2) partially stained; 3) fully stained blood vessels.