Deficiency in Lyst function leads to accumulation of secreted proteases and predisposition to mechanic stress-induced retinal detachment

Chediak–Higashi syndrome, caused by mutations in the Lysosome Trafficking Regulator (Lyst) gene, is a recessive hypopigmentation disorder characterized by albinism, neuropathies, neurodegeneration, and defective immune responses, with enlargement of lysosomes and lysosome-related organelles. Although recent studies have suggested that Lyst mutations impair the regulation of sizes of lysosome and lysosome-related organelle, the underlying pathogenic mechanism of Chediak–Higashi syndrome is still unclear. Here we show striking evidence that deficiency in LYST protein function leads to accumulation of photoreceptor outer segment phagosomes in retinal pigment epithelial cells, and reduces adhesion between photoreceptor outer segment and retinal pigment epithelial cells in a mouse model of Chediak–Higashi syndrome. In addition, we observe elevated levels of cathepsins, matrix metallopeptidase (MMP) 3 and oxidative stress markers in the retinal pigment epithelium of Lyst mutants. Previous reports showed that impaired degradation of photoreceptor outer segment phagosomes causes elevated oxidative stress, which could consequently lead to increases of cysteine cathepsins and MMPs in the extracellular matrix. Taken together, we conclude that the loss of LYST function causes accumulation of phagosomes in the retinal pigment epithelium and elevation of several extracellular matrix-remodeling proteases through oxidative stress, which may, in turn, reduce retinal adhesion. Our work reveals previously unreported pathogenic events in the retinal pigment epithelium caused by Lyst deficiency, which may place Chediak–Higashi syndrome patients at increased risk for retinal detachment. The same pathogenic events may be conserved in other professional phagocytic cells, such as macrophages in the immune system, contributing to overall Chediak–Higashi syndrome pathology.


INTRODUCTION
Chediak-Higashi syndrome (CHS) is a rare autosomal recessive disease characterized by albinism of the skin and hair, as well as hypopigmentation of the eye and additional eye pathologies including photophobia and macular hypoplasia associated with decreased visual acuity [1][2][3][4]. Patients also display progressive neurologic dysfunction, including motor and sensory neuropathies, ataxia, and progressive neurodegeneration [1,2,[5][6][7][8][9]. The most detrimental pathology is, however, recurrent bacterial infections, which predominantly affect the respiratory tract, skin, and mucous membranes. These infections are due to the dysfunction of polymorphonuclear leukocytes [1,2,10,11]. The majority of CHS cases progress to a lifethreatening lymphoproliferative accelerated phase characterized by massive hemophagocytic lymphohistiocytosis after exposure to Epstein-Barr virus [2]. Antibiotic treatments and hematopoietic stem cell transplantation have been used to combat recurrent infections and immunological complications; but these treatments target the symptoms, not the underlying pathogenic mechanism(s) [1,2,12].
CHS is caused by mutations in the ubiquitously expressed Lysosome Trafficking Regulator (Lyst) gene, which encodes LYST, a Beige and Chediak-Higashi (BEACH) domain-containing protein [12][13][14][15]. The loss of LYST function results in enlarged lysosomes and lysosome-related organelles (LROs) in all cell types examined [1,14,[16][17][18][19][20][21][22]. Functional studies using several model organisms have previously led to two distinct hypotheses for the effects of LYST in the regulation of LRO sizes: LYST may restrict homotypic lysosome fusion [23][24][25][26][27][28] by inhibiting membrane docking and fusion [23], or alternatively, LYST may promote lysosome fission [17,29,30]. However, a recent report indicates that LYST function is likely to be far more complex than a simple role in either lysosomal fusion or fission, and suggests that LYST may regulate fusion through fission-mediated recycling of the fusion machinery during lysosomal maturation [31]. Despite years of research, the exact molecular function of LYST remains unclear. Given that LYST is an extremely large protein, approximately 430 kDa in size, and contains multiple WD40 domains implicated in protein-protein interaction, it is likely that LYST has many alternate functions that are dictated by its interaction with binding partners. Thus, loss of LYST function may cause various cellular defects that have yet to be elucidated.
In this study, we examined the downstream effects of LYST dysfunction on the cellular pathology of the retinal pigment epithelium (RPE). The RPE is a monolayer of post-mitotic polarized epithelial cells, situated between the photoreceptors and the choroid; it is the primary caretaker of photoreceptor health and function [32]. One of the primary functions of the RPE is to engulf and degrade the distal tips of photoreceptor outer segments (POSs) [33,34]. Because each RPE cell serves many photoreceptor cells (200 in the mouse central retina) [35], they are tasked with degrading extraordinary amounts of POS material on a daily cycle [36,37].
By taking advantage of a novel mutant mouse strain bearing a mutation in Lyst, Lyst bg-18J . We found that, in addition to enlarged and redistributed lysosomes and accumulation of phagosomes in RPE cells, there is reduced adhesion between the RPE and the neural retina, causing higher risk in retinal detachment. We show that the accumulation of phagosomes is associated with increased oxidative stress and an elevation of a group of proteases, including cathepsins B, L, and S, and matrix metalloprotease 3 (MMP3), which are likely secreted into the interphotoreceptor matrix (IPM) and may contribute to the retinal adhesion defect. The disturbances in lysosomal enzymatic activities may have a profound impact on tissue integrity beyond the RPE. The same mechanism is likely to also exist in the immune system, where elevation of secreted proteases cleave cell surface proteins potentially leads to the reduced immune response observed in CHS [38]. Our results establish a series of events for the pathogenic mechanism of CHS and suggests an important new entry point for therapeutic intervention.

Mice
The Lyst mutant bg-18 (Lyst bg-18J ) was first identified by the JAX Mouse Mutant Resource as a spontaneous mutation, nm2144. Experimental animals were housed in the same mouse room and under the same 14-hr light / 10-hr dark cycle from birth. The light cycle was 6AM to 8PM. All experiments were approved by the Institutional Animal Care and Use Committee and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Indirect Ophthalmoscopy, Optical Coherence Tomography and Electroretinography
Indirect ophthalmoscopy, optical coherence tomography (OCT) and electroretinography (ERG) were performed as described [39]. For indirect ophthalmoscopy and OCT, 11 10-week-old wild type C57BL/6J male and 2 female mice, and 4 17-week-old mutant male mice, and 11 19-monthold male mice were used. Three wild type and 3 mutant 2-month-old male mice were used for ERG.

RNA preparation and reverse transcription
After carbon dioxide (CO 2 )-induced euthanasia, mouse eyes were dissected in DEPC-treated water. The posterior eyecup was separated from connective tissues and the iris epithelium, cornea, and lens. For RPE only RNA preparations, the RPE was peeled from the neuroretina.
RPE from both eyes of each animal was pooled. Poly A+ RNA from the RPE was extracted using Dynabead mRNA DIRECT Micro Kit (Invitrogen) according to the manufacturer's instructions. RNA concentration was quantified using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies). cDNA was synthesized using the RETROscript Kit (ThermoFisher Scientific).

Quantitative real-time PCR
Real-time PCR was performed using Bio-Rad iTaq mixture on Bio-Rad iCycler 96 thermocycler equipped with a CCD image detector, using protocols with a melt curve analysis. Only primers generating a solid major peak without obvious minor peaks in the melting curve were used.
Samples were collected from three wild type and three homozygous bg-18 mice. Each sample was subjected to three technical replications. Primers, PCR procedure and data analysis are described in detail in Supporting Information. PCR products were resolved by Metaphor agarose gel electrophoresis to confirm the expected sizes.

Quantification of melanin
After lysis of tissue samples with RIPA buffer, melanosomes precipitate in the insoluble pellet.
The insoluble pellet was dissolved in 1 N NaOH at 80˚C for 2 hr. Absorbance of commercial melanin pigment (Sigma) at defined concentrations was measured at 405nm to establish a standard curve. The melanin content of each sample was assessed at O.D. 405 and compared to the standard curve. The pigment concentration was normalized to the total protein concentration within each sample (μg melanin / mg protein).

Retinal adhesion assay
Retinal adhesion assays were performed at two time points, 9:00 AM (3 hours after onset of the light cycle) and 3:00 PM (9 hours after onset of the light cycle). Three wild type and three mutant 7-week-old mice were used for each time point. Enucleated eyes were submerged in 20 mM HEPES-buffered Hanks' saline solution containing calcium and magnesium (Mediatech, Inc., A Corning Subsidiary) at room temperature to preserve retinal adhesion. Eyecups were dissected as described above. A single radial cut toward the optic nerve was made to flatten the eyecups. The neural retina was slowly peeled from the underlying RPE and sclera with forceps from one side of the cut edge to the other. The peeled-off neural retina was then flattened, facing upward on a glass slide for imaging.
The imaging results were confirmed with western blot analysis, which measured the amount of RPE-specific proteins in the peeled-off neural retina. Blots were incubated with primary antibodies against ezrin (Cell Signaling, 3145, rabbit polyclonal, 1:1000), GFAP, and GAPDH.
Ezrin, a marker for RPE apical microvilli, was used to detect the amount of RPE attached to neural retina after separation from the RPE. GFAP and GAPDH were used as loading controls.

Transmission electron microscopy
Mice were perfused intracardially with buffered 1.2% (wt/vol) paraformaldehyde and 0.8% glutaraldehyde, and eyecups were processed using a standard procedure as described previously [40]. The eyes for ultrastructural analyses were fixed in an ice-cold fixative solution for 3 h. The anterior segment was removed and the posterior segment cut into 1 × 2 mm blocks. Additional fixation with 0.25% glutaraldehyde/0.2% paraform-aldehyde fixative was performed for 2-8 h followed by post-fixation with 1% osmium tetroxide. The dehydrated blocks were embedded in plastic. Tissue sections were cut and stained with uranyl acetate and lead citrate and examined with a JEM-1230 transmission electron microscope (JEOL, Ltd).

Immunostaining of RPE flat mounts
Enucleated eyes were submerged in ice-cold 1X PBS and dissected as described above. The For image analysis, only rhodopsin-positive POS fragments with diameter 0.5 µm-2.5 µm were counted by using the Imaris 9.1 software (Bitplane USA, Concord, MA, USA), with the aid of the 'Surfaces' rendering tool. Clustered POS fragments whose boundaries were difficult to delineate or those that clearly appeared to be cell surface adherent POS were not included in counts. The results were quantified and analyzed using unpaired t-test with Prism 6 software.

Pathogenic effects of Lyst deficiency in the RPE
We have identified a new mutant mouse bg-18 (B6.Cg-Lyst bg-18J /Boc, JR#028230, The Jackson Laboratory) that recapitulates many of the pathologies observed in human CHS patients and resembles the beige (Lyst bg-J ) mutant previously reported [22]. Homozygous bg-18 mutant mice, like beige mutant mice, display a dark grey coat color on the C57BL/6J background, compared with wild type littermates (Fig. 1A). The underlying mutation was first inferred as a mutation in Lyst in a whole exome sequencing project on 91 mutant mice [41]. Subsequent genome mapping, complementation test with the beige mutant, and sequencing of Lyst cDNA and genomic DNA from the bg-18 mutant definitively placed the causal mutation in the Lyst locus, as an intronic mutation causing a skipping of exon 10 resulting in a frameshift of the mRNA and truncation of the encoded LYST protein (Fig. S1). Details describing the genetic studies are included in the Supporting Text. We will refer to the bg-18J mutants as Lyst mutant mice, hereafter. This mutant was used to study the pathogenic effects of Lyst deficiency in this report.
The fundus images of homozygous Lyst mutants show an uneven distribution of retinal pigmentation compared to those of wild type mice (Fig. 1B, C). Optical coherence tomography (OCT) reveals two hyper-reflective layers corresponding to the RPE and choroid in the wild type retina, while only one layer is observed at the same location in the mutant retina (Fig. 1D, E), suggesting an alteration of the posterior retina.
By using transmission electron microscopy (TEM), we found that melanosomes in the RPE were remarkably enlarged both in the RPE and choroid of the Lyst mutants, relative to melanosomes in controls (Fig. 1F, G). Because the melanosomes tended to aggregate without distinct boundaries (Fig. S2), we were unable to distinguish individual organelles and quantitate their number.
However, a significant reduction of melanin concentration was observed in the Lyst mutant eyecups compared with wild type controls (Fig. 1H), indicating a reduction in melanogenesis or an increase in melanin turnover.
At 2 months of age, Lyst mutants show similar rod and cone electroretinography (ERG) responses to wild type controls (Fig. S3), indicating no obvious electrophysiological defects or retinal degeneration at this age. ERG measurements were also attempted on aged Lyst mutants but were not successful as their eyes could not be dilated. However, retinal degeneration was not observed at 19 months of age in OCT images of Lyst mutant eyes (Fig. S4).

Increased number of phagosomes in the Lyst mutant RPE
The tips of rod OS are phagocytosed by the RPE immediately after light onset [36,37]. We counted the numbers of phagosomes containing rod outer segments (OS) in RPE flat mounts from wild type and Lyst mutant mice at 7 to 8 weeks of age. At the time of light onset, the number of RPE phagosomes in Lyst mutant RPE were slightly higher compared with wild type control ( Fig. 2A, D). We also compared the numbers of RPE rhodopsin-positive phagosomes in IHC images of 5-week-old wild type and Lyst mutant eyes at 0.5, 1, and 2 hr after light onset and found that the number of phagosomes were higher in the mutant RPE at all time points examined (Fig. S5). The most remarkable differences in the number of rhodopsin-positive phagosomes in Lyst mutant RPE cells when compared to wild type, however, was at 3 or 9 hr after light onset ( Fig. 2B, C, E, F, G; Fig. S6).
The increase in RPE phagosomes may either be due to an elevation of phagocytic activity or reduction in lysosomal degradation of engulfed phagosomes in the RPE. To differentiate between these two possibilities, the level of active, phosphorylated MER proto-oncogene tyrosine kinase (MERTK), which has been reported to increase upon the onset of phagocytosis [42], was assessed by western blot analysis. No obvious change was observed right after light onset (Fig. 2H, I), the time point of peak phagocytosis reported previously [43]. As MERTK is only localized on the apical surface of the RPE cells, this result indicates that phagocytic activity is not affected in the RPE of Lyst mutants. In addition, we found that the phosphorylated MERTK was still mainly observed in the Lyst mutant RPE at the time of lights-on but not at later sampled time points (Fig. S5). This finding indicates that phagocytosis of POS tips in LYSTdeficient RPE cells is not prolonged and suggests that the phagocytic activity and the duration of phagocytosis are not affected by the Lyst mutation. Thus, we propose that the degradation of engulfed phagosomes, by lysosomes, is impaired, which results in the accumulation of undigested phagosomes in the RPE. Previous research has shown that the fusion between phagosomes and lysosomes is blocked by the re-distribution of lysosomes to perinuclear clusters induced by lipopolysaccharide (LPS) treatment in dendritic cells [44]. Other studies have demonstrated giant and perinuclear lysosomes in Lyst-deficient cells [19,45,46]. Our Lyst bg-18J mice recapitulate this subcellular phenotype of lysosomes (Fig. S7). We postulate that the perinuclear aggregation of lysosomes contributes to the accumulation of phagosomes in the RPE of Lyst mutants by blocking phago-lysosomal fusion.

Reduced retinal adhesion in LYST deficient mice
When we performed histology, and compared the wild type ( Fig. 3A-C) to the 3-week-old Lyst mutant retinas (Fig. 3D), we observed small areas of detachment of the neural retina from the RPE in sections from every homozygous Lyst mutant mouse examined (n>3 for each time cohort). The detachment was observed as early as 1 month of age (Fig. 3E), which appeared as larger areas of detachment by 3 months of age (Fig. 3F). However, retinal detachment is not observed by OCT at 1 or 3 months (Fig. S8). The detachment observed by histology may result from a meaningful artifact that occurs during enucleation of the eye and histological sample preparation due to the reduced retinal adhesion and is, therefore, not observed by OCT, which is an in situ technique. This suggests that the Lyst mutation results in the reduction of retinal adhesion, which only manifests as a full-blown detachment when the retina is stressed mechanically.
Separation of the neural retina from its supportive structures can lead to significant visual defects [47][48][49][50][51][52]. The microvilli-rich apical domains of RPE cells, which form the outermost layer of the retina, interdigitate with the photoreceptor outer segments. Thus, apical domains of RPE cells or even entire RPE cells may remain attached when the neural retina is mechanically peeled from the RPE [42]. To determine if the retinal detachment observed by histology, but not by OCT, was an artifact of histological sample preparation or the consequence of reduced retinal adhesion insufficient to induce detectable retinal detachment in vivo, we compared the adhesion between the RPE and neuroretina in wild type and Lyst mutants accordingly. In the wild type peeled-off neural retina, RPE pigment was abundantly attached to the surface of photoreceptor outer segments (POS); whereas the peeled-off neural retina of the age-matched Lyst mutant mice (separated under the same experimental conditions and at similar times of day), were almost devoid of RPE pigment (Fig. 4A, B). To quantify the alteration in adhesion, the level of ezrin, an RPE microvillus marker was measured by western blot analysis in the peeled-off retinas.
Compared to wild type neural retina, the ezrin levels were significantly reduced in the Lyst mutant peeled-off neural retina (Fig. 4C, D). In contrast, ezrin levels were similar in the wild type and mutant whole eyecups (Fig. 4E, F). Therefore, the reduction in ezrin levels in the mutant peeled-off neural retina is likely due to poor RPE-retina adhesion in the Lyst mutant mice, and not due to a decrease in ezrin levels in the RPE microvilli.

Levels of cysteine cathepsins and MMP3 are increased in the Lyst mutant eyecups
The POS and RPE microvilli of the Lyst mutant mice were not as densely packed as those in the wild type retina, but the structure of the processes were still clearly visible and comparable to that of the wild type mice (Fig. S9). In addition, we also examined ERM (ezrin, radixin, moesin) proteins that are activated upon phosphorylation (Thr567 of ezrin, Thr564 of radixin, Thr558 of moesin) and link membrane-associated proteins to actin filaments at the cell cortex, making them important for RPE microvilli morphology. No significant differences were observed in the level of phospho-ERM between wild type and Lyst eyecups (Fig. S9). These results suggest that the reduced adhesion between the neural retina and RPE in the mutant eye is unlikely to be due to structural alterations.
Adhesion of the neuroretina to the RPE is mediated by the interphotoreceptor matrix (IPM), a retina-specific type of extracellular matrix (ECM) between the RPE and POS. Previously, it was shown that proteases, such as matrix metalloproteinases (MMPs) and cysteine cathepsins, are found in the ECM and play important roles in ECM remodeling [53][54][55][56]. Quantitative real-time PCR showed up-regulation of transcripts of cathepsins and MMP12 in Lyst mutant ocular tissues [57]. To confirm previous reports and pinpoint the precise locations where the upregulations actually occur, we tested transcripts of cathepsin B, L, and S and MMP3 in the RPE, because it appears to be the most affected layer in the retina by the Lyst mutation. Consistent with previous reports, we also found that the transcription levels of MMP3, and cathepsin B, L, and S assessed by quantitative real-time PCR, are increased in the RPE of the Lyst mutants relative to control (Fig. 5A-D). However, in both wild type and the Lyst mutant RPE, MMP3 and cathepsin S transcriptional expression was very low, and the protein levels, assessed by western blot analysis, were below detection level. Thus, in this study, our analysis focused on cathepsin B, as it is the most abundant cysteine protease in the RPE [58]. However, other aforementioned proteases upregulated by Lyst deficiency may play similar roles in the IPM.
Cathepsin B is a lysosomal cysteine protease normally found ubiquitously in cells and tissues [59]. In malignant tumors, the expression of cathepsin B is highly upregulated and mature cathepsin B is secreted to the cell surface where it can degrade ECM proteins. The degradation of ECM proteins by cathepsin B is required for tumor cell invasion and metastasis [53,55,56,60]. To test if molecular changes of cathepsin B could be responsible for the reduced adhesion between RPE and neural retina in the Lyst mutant eye, cathepsin B protein levels were measured by western blot analysis. In the Lyst mutant eyecups, a significant elevation of mature cathepsin B protein was observed (Fig. 5E, F). Mature cathepsin B, which was mainly found in the RPE samples and not in neural retina samples, was dramatically increased relative to controls (Fig.  5G). By immunohistochemical staining, a strong cathepsin B signal was observed at the apical surface of the RPE, which colocalized with ezrin ( Fig. 5H-M), and juxtaposed to the rhodopsinlabeled POS (Fig. S10). Combined with the QPCR and western blot results, these results indicate that perhaps the elevated levels of mature cathepsin B on the apical surface of RPE and interfacing POS may contribute to the reduced retinal adhesion in the Lyst mutant mice.

The level of oxidative stress is increased in the Lyst mutant RPE
Previous studies have reported elevated expression levels of cysteine cathepsins in mouse RPE exposed to oxidative stress [58]. MMPs were also reported to be upregulated by oxidative stress in many other cell types [61][62][63][64]. These data suggest that the observed increase in levels of cysteine cathepsins and MMP3 of Lyst mutant RPE cells may be caused by elevated levels of oxidative stress as well. By western analysis, we determined that the level of 4-hydroxynonenal (4-HNE)-modified protein, a widely used biomarker for oxidative stress, was increased in the Lyst mutant RPE cells compared to wild type control (Fig. 6). Thus, we postulate that the oxidative stress due to Lyst deficiency results in elevation of secreted proteases in the IPM between the RPE and the photoreceptors.

DISCUSSION
In this study, our results demonstrate for the first time that LYST plays an important role in phagosome processing in RPE cells. We observed an abnormal accumulation of phagosomes in RPE cells in Lyst mutant mice ( Fig. 2 and Fig. S5). Previous studies have suggested that phagocytosis of photoreceptor outer segments by RPE cells causes oxidative stress. For example, long-term daily feeding of rod OS increases the number of intracellular autofluorescent granules and increases catalase activity in cultured human RPE cells [65,66], and rod OS uptake by cultured human RPE cells increases oxygen consumption and intracellular H 2 O 2 production [67].
Thus, we reasoned that the accumulation of phagosomes may cause an increase in oxidative stress in the Lyst mutant RPE cells. In support of this hypothesis, studies have shown that slowed degradation of POS phagosomes causes oxidative stress characterized by the increase of oxidative stress markers, i.e., malondialdehyde (MDA) or 4-HNE levels, in the RPE [68,69].
Our data shows that the levels of 4-HNE-modified proteins are increased in the Lyst mutant RPE compared with those in the wild type control samples, indicative of oxidative stress (Fig. 5).
Previous studies in tumor tissues suggested that oxidative stress could cause elevation of secreted cathepsin B and MMPs, which contributed to the digestion of extracellular matrix during tumor metastasis [53,55,56,60]. In addition, in a mouse model of chronic oxidative stress, termed Hyperoxia-Related Retinal Degeneration (HRRD), elevation of transcript and protein levels of cathepsin B, L, and S in the RPE have been reported [58]. Strikingly, we also observed an suggest that in Lyst mutant RPE, phagosome accumulation leads to oxidative stress, which results in increased expression of proteases on the RPE apical surface, thus reducing overall RPE-retinal adhesion. Although we observed visible retinal detachment only after mechanical stress during removal of the eye from the animal, it is known that blunt ocular trauma is a common cause for retinal detachment [70][71][72]. Our findings suggest that Chediak-Higashi patients may be at increased risk of retinal detachment after ocular trauma.
Oxidative stress contributes to the pathogenesis of many neurodegenerative diseases including age-related macular degeneration (AMD). In addition, cathepsins have been implicated in the pathogenesis of AMD [58,[73][74][75]. Whether oxidative stress contributes to AMD via aberrant activation of cathepsins is still unclear. Research on the regulation of cysteine cathepsins in the context of oxidative stress may provide new therapeutic targets for AMD. Thus, the Lyst mutant mice may be a valuable genetic model to study the impact of chronic oxidative stress on the RPE. RPE cells are highly phagocytic, and thus examining this cell type provides a unique opportunity to study phagocytosis in vivo. A similar pathological pathway described here may also exist in Lyst-deficient leukocytes where excessive secreted cathepsin B and other secreted proteases may cleave surface antigen or receptors, thus contributing to the decreased immune response seen in CHS patients. Alternatively, Lyst deficiency may primarily reduce the immune response through decreased digestion of phagosomes by lysosomes in leukocytes, which is critical for antigen presentation. The exact outcome caused by phagosome accumulation-induced oxidative stress in leukocytes versus cells in tissue such as the RPE will have to be determined experimentally in the future.

FUNDING
This work is supported by NIH Grant EY011996 and EY027860 to P.M.N., and EY019943 to B.C. The Jackson Laboratory Scientific Services is supported by NIH Grant CA034196. Partial funding from R01EY027442 and core grant P30EY000331 (to D.S.W.) is acknowledged. Figures   Fig. 1