Abstract

Background and objectives

   Fibrosis contributes to 45% of deaths in industrialized nations and is
   characterized by an abnormal accumulation of extracellular matrix
   (ECM). There are no specific anti-fibrotic treatments for liver
   fibrosis, and previous unsuccessful attempts at drug development have
   focused on preventing ECM deposition. Because liver fibrosis is largely
   acknowledged to be reversible, regulating fibrosis resolution could
   offer novel therapeutical options. However, little is known about the
   mechanisms controlling ECM remodeling during resolution. Changes in
   proteolytic activity are essential for ECM homeostasis and macrophages
   are an important source of proteases. Herein, in this study we evaluate
   the role of macrophage-derived cathepsin D (CtsD) during liver
   fibrosis.

Methods

   CtsD expression and associated pathways were characterized in
   single-cell RNA sequencing and transcriptomic datasets in human
   cirrhosis. Liver fibrosis progression, reversion and functional
   characterization were assessed in novel myeloid-CtsD and
   hepatocyte-CtsD knock-out mice.

Results

   Analysis of single-cell RNA sequencing datasets demonstrated CtsD was
   expressed in macrophages and hepatocytes in human cirrhosis. Liver
   fibrosis progression, reversion and functional characterization were
   assessed in novel myeloid-CtsD (CtsD^ΔMyel) and hepatocyte-CtsD
   knock-out mice. CtsD deletion in macrophages, but not in hepatocytes,
   resulted in enhanced liver fibrosis. Both inflammatory and matrisome
   proteomic signatures were enriched in fibrotic CtsD^ΔMyel livers.
   Besides, CtsD^ΔMyel liver macrophages displayed functional,
   phenotypical and secretomic changes, which resulted in a degradomic
   phenotypical shift, responsible for the defective proteolytic
   processing of collagen I in vitro and impaired collagen remodeling
   during fibrosis resolution in vivo. Finally, CtsD-expressing
   mononuclear phagocytes of cirrhotic human livers were enriched in
   lysosomal and ECM degradative signaling pathways.

Conclusions

   Our work describes for the first-time CtsD-driven lysosomal activity as
   a central hub for restorative macrophage function during fibrosis
   resolution and opens new avenues to explore their degradome landscape
   to inform drug development.

   Keywords: Fibrosis, Protease, Cathepsin, Resolution, Macrophage

Graphical abstract

   [63]Image 1
   [64]Open in a new tab

Highlights

     * •
       Human cirrhotic CtsD-expressing macrophages are enriched in ECM
       degradative pathways.
     * •
       Myeloid-CtsD KO mice present enhanced liver fibrosis and impaired
       resolution.
     * •
       CtsD contributes to functional, phenotypical and secretomic changes
       in macrophages.
     * •
       CtsD-driven macrophage degradomic shift is essential for fibrosis
       resolution.

1. Introduction

   Liver fibrosis is the result of a dysregulation of the normal
   wound-healing response due to repetitive damage or chronic injury.
   During liver fibrosis, replacement of the functional parenchyma by
   excessive electrodense extracellular matrix (ECM) results in
   architectural remodeling and a progressive stiffening of the liver,
   which can eventually lead to organ failure. It is currently estimated
   that fibrosis is a contributing factor to 45% of deaths in
   industrialized nations [[65]1] and it exponentially increases
   liver-related mortality in nonalcoholic fatty liver disease patients
   [[66]2]. Despite liver fibrosis being widely accepted to be reversible,
   little is known about the biological mechanisms controlling ECM
   remodeling and reabsorption. Changes in proteolytic activity are
   essential for ECM homeostasis, however, the proteolytic network
   involved in ECM deposition and reabsorption is far from understood.
   Considering that no specific treatments for liver fibrosis are
   currently available and that previous research efforts designing
   anti-fibrotic drugs focused on stopping ECM deposition have failed,
   increasing our biological understanding about the proteolytic pathways
   controlling ECM reabsorption can provide novel therapeutic approaches
   for the treatment of fibrotic diseases.

   Cathepsins (Cts) are peptide-bond cleaving enzymes classically referred
   as lysosomal proteases. However, it is recognized that Cts can be found
   in other cellular locations and display activity outside their optimal
   acidic pH range [[67]3]. They are classified into three families
   depending on their structure and catalytic sites: serine (CtsA and G),
   aspartate (CtsD and E) and cysteine (CtsB, C/DPP1, F, H, K, L, O, S, V,
   W and Z/X). Proteases finely tune several biological processes through
   efficient, highly selective, and limited cleavage of specific
   substrates, a process also known as proteolysis. Growing evidence
   suggests important roles for mainly cysteine Cts in the context of
   liver disease, participating in apoptosis, activation of hepatic
   stellate cells (HSC), autophagy, inflammation, matrix remodeling, lipid
   metabolism and cancer. However, our understanding of their specific
   targets and signaling networks remains very limited [[68]3]. CtsB is
   one of the best studied Cts in the context of liver fibrosis, and it
   appears to have a dual role acting as a pro-apoptotic signal for
   hepatocytes [[69]4] and a pro-fibrogenic signal for HSC [[70][5],
   [71][6], [72][7]]. Contrary to CtsB, the study of aspartyl CtsD has
   been challenging, partially due to the lack of an adult CtsD global
   knock-out mouse [[73]8]. Recently, the generation of a CtsD floxed
   mouse [[74]9] has opened new avenues to study the cell-specific role of
   CtsD during disease progression. Therefore, the current study was
   designed to evaluate CtsD cell-specific contribution to liver fibrosis.
   Our analysis of cirrhotic human tissues revealed high expression of
   CtsD in liver macrophages and hepatocytes, which was confirmed by the
   analysis of publicly available human single-cell RNA sequencing
   datasets. Hence, we generated two novel CtsD knock-out mice for myeloid
   cells (CtsD^ΔMyel) and hepatocytes (CtsD^ΔHep) to explore the
   contribution of CtsD expression in these cell types to liver fibrosis
   development. CtsD deletion in myeloid cells, but not in hepatocytes,
   resulted in increased liver fibrosis. In addition, proteomic analysis
   of fibrotic livers demonstrated an enrichment of the immune system and
   the matrisome signaling pathways in fibrotic livers from CtsD^ΔMyel
   mice. CtsD^ΔMyel liver macrophages displayed functional, phenotypical
   and secretomic changes that resulted in defective lysosomal proteolytic
   processing of collagen I in vitro and impaired collagen remodeling
   during fibrosis resolution in vivo. Analysis of CtsD mononuclear
   phagocyte (MP) subpopulations in cirrhotic human livers confirmed that
   CtsD-expressing subpopulations were differentially enriched in
   lysosomal and ECM containing and degradative signaling pathways. Our
   study demonstrates that CtsD is essential in regulating CtsD-driven
   lysosomal activity of macrophages, influencing their phenotype,
   secretomic profile and collagenolytic activity during liver fibrosis
   progression and resolution.

2. Methods

2.1. Generation of CtsD knock-out mouse strains for macrophages (CtsD^ΔMyel)
and hepatocytes (CtsD^ΔHep)

   To obtain the conditional KO mice for CtsD in myeloid cells or
   hepatocytes, LysMCre mice (Jackson Laboratory
   B6.129P2-Lyz2tm1(cre)Ifo/J) [[75]10] or AlbCre (Jackson Laboratory
   B6.Cg-Tg(Alb-cre)21Mgn/J) [[76]11] were bred with CtsD floxed mice
   [[77]9]. Briefly, CtsD^F/F LysMCre^−/− or CtsD^F/F AlbCre^−/− females
   were bred with CtsD^F/F LysMCre^+/− or CtsD^F/F AlbCre^+/− males to
   generate CtsD^F/F-LysMCre^+/− (CtsD^ΔMyel) and CtsD^F/F-LysMCre^−/−
   (CtsD^F/F) or CtsD^F/F-AlbCre^+/− (CtsD^ΔHep) and CtsD^F/F-AlbCre^−/−
   (CtsD^F/F) littermates, respectively. CtsD^ΔMyel and CtsD^ΔHep were
   bred in heterozygosis and kept as separate strains. Genotyping was
   performed by PCR to detect the cre allele and floxed alleles in ear
   clipped biopsies and cre recombined allele in liver macrophages for the
   CtsD^ΔMyel or hepatocytes for the CtsD^ΔHep. Randomized littermates
   were used for the studies. Samples obtained from the animal studies
   were coded and histological analysis and scoring were performed blind.
   Males were preferentially used for the experimental models unless
   stated otherwise, as cathepsin D gene expression can be controlled by
   estrogens [[78]12,[79]13]. Females were included in all cell isolation
   experiments, as we did not detect any differences between male and
   female liver macrophages, and in the later in vivo experimental model
   of liver fibrosis reversion. All animal studies were done in accordance
   with the Comitè Ètic d'Experimentació Animal (CEEA-UB), under the
   approval of the Generalitat de Catalunya and according to Animal
   Research: Reporting of In Vivo Experiments (ARRIVE) guidelines. Primer
   sequences for genotyping are detailed in [80]Table S1.

2.2. Liver fibrosis experimental animal models

   8–12 weeks old males were preferentially used for the experimental
   models unless stated otherwise, as cathepsin D gene expression can be
   controlled by estrogens [[81]12,[82]13]. Hence, this can be translated
   into a higher variability of CtsD expression in females depending on
   their level of estrogens.

   For the short-term chronic carbon tetrachloride (CCl[4]) model,
   CtsD^ΔMyel and CtsD^F/F or CtsD^ΔHep and CtsD^F/F littermates were
   injected intraperitoneally (IP) twice a week with 2 μL (CCl[4]/olive
   oil, 1:3 [v/v])/g body weight or olive oil for 8 weeks. For the bile
   duct ligation model, the bile duct was ligated for 7 or 14 days in
   CtsD^ΔMyel and CtsD^F/F or for 14 days in CtsD^ΔHep and CtsD^F/F mice.
   A sham operation was performed as control group. For the long-term
   chronic carbon tetrachloride (CCl[4]) model, CtsD^ΔMyel or CtsD^F/F
   mice were injected intraperitoneally once a week with 2 μL
   (CCl[4]/olive oil, 1:10 [v/v])/g body weight or olive oil for 24 weeks.
   At least 6 animals were used per experimental group.

2.3. Liver reversion experimental animal model

   Fibrosis was established for 4 weeks by intraperitoneal administration
   of 2 μL (CCl[4]/olive oil, 1:3 [v/v])/g body weight twice a week in
   CtsD^ΔMyel and CtsD^F/F male and female mice. Endpoints were 24 h and
   72 h after the last CCl[4] dose for the peak and the resolution
   experimental groups, respectively.

2.4. Statistical analysis

   Results are expressed as mean ± SEM unless otherwise stated in the
   figure legend. All p-values were calculated using 1 way ANOVA followed
   by Bonferroni's post-test, 2 way ANOVA followed by Turkey's or
   Bonferroni's post-test or two tailed unpaired student's t-test.
   ^∗P ≤ 0.05; ^∗∗P ≤ 0.01; ^∗∗∗P ≤ 0.001 was considered statistically
   significant.

   More details about the experimental procedures can be found in the
   supplementary material.

3. Results

3.1. CtsD expression increases in macrophages and serum from cirrhotic
patients

   First, we investigated CtsD expression in publicly available
   single-cell RNA sequencing (scRNAseq) datasets from five normal and
   five cirrhotic livers [[83]14]. Our analysis identified 11 common cell
   lineages, annotated according to the original publication ([84]Fig. S1
   and [85]Table S2), across control and cirrhotic livers. From those,
   CtsD was predominantly expressed in the mononuclear phagocyte
   population in both control ([86]Figure 1A) and cirrhotic
   ([87]Figure 1B) livers, but it was also expressed to a lesser extend in
   endothelial and epithelial cells. Interestingly, while the percentage
   of CtsD-expressing cells increased in the cirrhotic epithelial cells,
   it decreased in cirrhotic endothelial cells. Next, we analyzed the
   expression of CtsD in tissue biopsies from control and cirrhotic
   patients. Immunofluorescent staining confirmed that CtsD was expressed
   in both macrophages (CD68+ve) and hepatocytes in control and cirrhotic
   tissues ([88]Figure 1C), displaying a common cytosolic vesiculated
   pattern compatible with lysosomal expression, in agreement with
   classical studies performed by Yokota and co-workers in rat liver
   [[89]15]. Further transcriptomic analysis of KCs isolated from control
   and cirrhotic human livers revealed a significant up-regulation of CtsD
   in human cirrhotic KCs (Log2 Fc = 2.63 p ≤ 0.001) [[90]16]. Finally, we
   observed that CtsD activity was also significantly increased in serum
   from cirrhotic patients ([91]Figure 1D).

Figure 1.

   [92]Figure 1
   [93]Open in a new tab

   CtsD is elevated in human cirrhosis and is highly expressed in liver
   MPs.

   UMAP projection depicting CtsD expression across cell types and
   dotplots representing percentage and average expression of CtsD within
   human (A) control and (B) cirrhotic livers ([94]GSE136103). (C)
   Representative images of dual IF staining for CtsD (green) and CD68
   (red) in human control and cirrhotic livers. (D) CtsD activity in serum
   from control and cirrhotic patients. Scale bars represent 20 μm. T-test
   analysis, ∗∗p ≤ 0.01.

3.2. Characterization of CtsD^ΔMyel and CtsD^ΔHep mice reveals normal basal
phenotype

   Our results demonstrated that macrophages and, to a lesser extent,
   hepatocytes, are the main cell types expressing CtsD in human
   cirrhosis. Thus, to better understand the role of CtsD in fibrosis
   development we generated two novel CtsD cell-specific knock-out mice
   for myeloid lineage cells including liver macrophages, both resident
   [[95]17] and infiltrated [[96]18], or for hepatocytes by breeding CtsD
   floxed mouse [[97]9] with LysMCre [[98]10] or AlbCre [[99]11] mice
   respectively, from now on referred as CtsD^ΔMyel or CtsD^ΔHep. To
   validate CtsD floxed allele recombination in liver macrophages, we
   assessed CtsD expression by WB in liver macrophages isolated from
   CtsD^F/F and CtsD^ΔMyel mice, demonstrating a significant decrease of
   CtsD in F4/80+ve cells from CtsD^ΔMyel mice ([100]Figure 2A). In
   agreement with results obtained from the human cirrhotic liver
   macrophages, CtsD^F/F liver macrophages isolated from CCl[4] injured
   livers displayed a significant increase in CtsD gene expression
   ([101]Figure 2B) and activity ([102]Figure 2C) that was absent in
   CtsD^ΔMyel liver macrophages. Validation of CtsD^ΔHep mouse was carried
   out by WB in primary mouse hepatocytes isolated from CtsD^F/F and
   CtsD^ΔHep mice ([103]Figure 2D), demonstrating a lack in CtsD
   expression in hepatocytes from CtsD^ΔHep mice. Of note, no significant
   differences in the gene expression of other cathepsins (CtsA, B, C, L,
   S, X/Z) between phenotypes were observed in liver macrophages nor in
   hepatocytes ([104]Figs. S2A–B). Similarly to what was observed in human
   cirrhotic livers, mouse CtsD^F/F fibrotic livers displayed the
   strongest CtsD signal in macrophages (F4/80+ve cells) and a mild signal
   in hepatocytes with a cytosolic vesiculated pattern compatible with
   lysosomes. As expected, CtsD signal in macrophages or in hepatocytes
   from CtsD^ΔMyel or CtsD^ΔHep livers respectively was not detected
   ([105]Figure 2E). Next, analysis of serum alanine aminotransferase
   (ALT) and liver histology revealed no liver damage and normal liver
   histology in both CtsD^ΔMyel and CtsD^ΔHep adult healthy mice
   ([106]Figure 2F) and no differences in the liver to body weight ratios
   ([107]Figure 2G). In addition, organ to body weight ratios and
   histology of extrahepatic tissues (brain, spleen, intestine and
   kidneys) from adult healthy CtsD^ΔMyel mouse, which are affected by the
   widespread deletion of CtsD in the global CtsD KO mouse [[108]8],
   demonstrated no affectation in comparison with CtsD^F/F mouse
   ([109]Figure 2H–I).

Figure 2.

   [110]Figure 2
   [111]Open in a new tab

   Characterization and validation of CtsD^ΔMyeland CtsD^ΔHepmice.

   (A) F4/80, mature CtsD and β-actin expression by WB from liver
   macrophages of CtsD^F/F and CtsD^ΔMyel mice and WB quantification.
   (B)CtsD gene expression of liver macrophages isolated from CtsD^F/F and
   CtsD^ΔMyel mice treated ± CCl[4] for 72 h and (C) CtsD activity in
   liver macrophages isolated from CtsD^F/F and CtsD^ΔMyel mice treated
   with CCl[4] for 72 h. (D) Pro-CtsD, mature CtsD and β-actin expression
   by WB of hepatocytes from CtsD^F/F and CtsD^ΔHep mice and WB
   quantification. (E) CtsD (green) and F4/80 (red) dual IF in fibrotic
   liver tissue from CtsD^F/F, CtsD^ΔMyel and CtsD^ΔHep mice. (F) Serum
   ALT and liver H&E representative images from CtsD^F/F, CtsD^ΔMyel and
   CtsD^ΔHep mice. (G) Liver to body weight ratio in CtsD^F/F, CtsD^ΔMyel
   and CtsD^ΔHep mice. (H) Organ to body weight ratios from brain (BR),
   thymus (TH), spleen, (SPL) and kidney (KD) and (I) representative
   images of brain, spleen, intestine and kidney tissues from 10-week-old
   CtsD^F/F and CtsD^ΔMyel mice stained with Cresyl violet for the brain,
   H&E for the spleen and intestine and PAS for the kidney. Scales
   represent 20 μm in panel E and 50 μm in panel F and I. Data in A and E
   is expressed as percentage of CtsD signal versus β-actin. 1 way ANOVA
   or T-test analysis, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001.

3.3. Liver fibrosis is exacerbated in CtsD^ΔMyel mouse, but not in CtsD^ΔHep
one

   Next, we investigated liver fibrosis development in CtsD^ΔMyel or
   CtsD^ΔHep in comparison with CtsD^F/F mice. To induce liver fibrosis,
   both CtsD^F/F and CtsD^ΔMyel or CtsD^F/F and CtsD^ΔHep littermates were
   injected with CCl[4] or Olive Oil, as study controls, twice a week for
   8 weeks as depicted in [112]Figure 3A or [113]Figure 3J. As expected,
   CtsD IHC demonstrated lack of CtsD in tissue macrophages from
   OOil-CtsD^ΔMyel and 8wkCCl[4]-CtsD^ΔMyel livers ([114]Figure 3B) or in
   hepatocytes from OOil-CtsD^ΔHep and 8wkCCl[4]-CtsD^ΔHep livers
   ([115]Figure 3K). When studying matrix deposition, liver collagen
   content was increased in all CCl[4]-treated groups, however, only
   8wkCCl[4]CtsD^ΔMyel but not 8wkCCl[4]CtsD^ΔHep, displayed significantly
   enhanced collagen deposition as determined by Sirius Red morphometric
   analysis of tissue liver sections ([116]Figure 3C,L) and liver
   hydroxyproline content ([117]Figure 3D,M). Classical fibrogenic markers
   Col1A1 and TGF-β were assessed by RT-qPCR and WB, last one only for
   TGF-β ([118]Figure 3E–G), and α-SMA expression was determined by IHC
   ([119]Figure 3H) in liver tissue. All of them were significantly
   increased only in 8wkCCl[4]-CtsD^ΔMyel livers compared to
   8wkCCl[4]-CtsD^F/Fones, but not in 8wkCCl[4]-CtsD^ΔHep ([120]Figures 3N
   and [121]S3A-B). Of note, liver damage (serum ALT) was increased after
   CCl[4] treatment but no significant differences were observed between
   phenotypes ([122]Figures 3I and [123]S3C) after CCl[4] administration.
   Prolonged 6-month administration of CCl[4], at a lower dose
   ([124]Figure S3D), resulted in comparable results to those obtained
   with the 8 weeks CCl[4] model, demonstrating increased liver collagen
   accumulation and α-SMA expression in 6MCCl[4]-CtsD^ΔMyel compared to
   6MCCl[4]-CtsD^F/F mice ([125]Fig. S3E–H).

Figure 3.

   [126]Figure 3
   [127]Open in a new tab

   CtsD^ΔMyelmice, but not CtsD^ΔHep, display enhanced liver fibrosis
   after 8-weeks CCl[4]administration.

   (A) Experimental design of the 8-weeks CCl[4] study in CtsD^ΔMyel mice.
   (B) Representative liver tissue images of CtsD IHC (black arrows
   indicate CtsD+ve macrophages), (C) morphometric analysis and
   representative images of Sirius Red (SR)+ve area/field (D) liver
   hydroxyproline quantification, liver (E)Col1A1 and (F)TGF-β gene
   expression from 8-weeks Olive Oil (OOil) or CCl[4] treated CtsD^F/F and
   CtsD^ΔMyel mice. (G) WB for TGF-β and β-actin in liver lysates from
   8wkCCl[4]-CtsD^F/F and 8wkCCl[4]-CtsD^ΔMyel mice and WB quantification.
   (H) Morphometric analysis of α-SMA+ve area/field with representative
   images with zoomed area and (I) serum alanine aminotransferase from
   8-weeks Olive Oil (OOil) or CCl[4] treated CtsD^F/F and CtsD^ΔMyel
   mice. (J) Experimental design of the 8-weeks CCl[4] study in CtsD^ΔHep
   mice. (K) Representative liver tissue images of CtsD IHC; (L)
   morphometric analysis and representative images of Sirius Red (SR)+ve
   area/field; (M) liver hydroxyproline quantification and (N) liver
   Col1A1 gene expression from 8 weeks Olive Oil (OOil) or CCl[4] treated
   CtsD^F/F and CtsD^ΔHep mice. Scale bars represent 50 μm. 1-way ANOVA,
   ∗p ≤ 0.05, ∗∗p ≤ 0.01 or ∗∗∗p ≤ 0.001.

   To determine whether the effect of CtsD deletion in macrophages or
   hepatocytes on fibrosis was common to different types of injury, we
   performed bile duct ligation (BDL) or sham surgery in CtsD^F/F and
   CtsD^ΔMyel mice ([128]Figure 4A) or CtsD^F/F and CtsD^ΔHep mice
   ([129]Figure 4I) for 14 days. CtsD deletion in macrophages and
   hepatocytes was confirmed by IHC in liver tissue from CtsD^ΔMyel or
   CtsD^ΔHep mice, respectively ([130]Figure 4B,J). In agreement with the
   8wk-CCl[4] model, BDL-CtsD^ΔMyel mice displayed increased liver
   collagen deposition ([131]Figure 4C–D) but no differences were observed
   in BDL-CtsD^ΔHep mice when comparing with their BDL-CtsD^F/F
   littermates ([132]Figures 4K–M and S4A,B). Unlike the previous study,
   no significant differences were detected in Col1A1, TGF-β or α-SMA
   expression between CtsD^ΔMyel and CtsD^F/F after 14 days of BDL
   ([133]Figure 4E–G). Nevertheless, early changes in gene expression
   showing increased Col1A1, TGF-β and ACTA2 were demonstrated 7 days
   after BDL in CtsD^ΔMyel compared to CtsD^F/F ([134]Figs. S4D–G).
   Finally, as in the 8wk-CCl[4] model, no significant differences were
   observed in liver damage after BDL amongst the different phenotypes
   studied ([135]Figures 4H, [136]S4C and S4H). Our results demonstrate
   that CtsD deletion in macrophages, but not in hepatocytes, contributes
   to an enhancement in liver fibrosis in two etiologically different
   models of experimental fibrosis without inducing changes in liver
   damage.

Figure 4.

   [137]Figure 4
   [138]Open in a new tab

   CtsD deletion in macrophages, but not in hepatocytes, exacerbates liver
   fibrosis after BDL.

   (A) Experimental design of the 14-day BDL study in CtsD^ΔMyel mice. (B)
   Representative liver tissue images of CtsD IHC (black arrows indicate
   CtsD+ve macrophages), (C) morphometric analysis and representative
   images of Sirius Red (SR)+ve area/field (D) liver hydroxyproline
   quantification, liver (E)Col1A1 and (F)TGF-β gene expression, (G)
   morphometric analysis of α-SMA+ve area/field and representative images
   with zoomed area and (H) serum alanine aminotransferase from 14 days
   Sham or BDL CtsD^F/F and CtsD^ΔMyel mice. (I) Experimental design of
   the 14-day BDL study in CtsD^ΔHep mice. (J) Representative liver tissue
   images of CtsD IHC; (K) morphometric analysis and representative images
   of Sirius Red (SR)+ve area/field; (L) liver hydroxyproline
   quantification and (M) liver Col1A1 gene expression from 14 days Sham
   or BDL CtsD^F/F and CtsD^ΔHep mice. Scale bars represent 50 μm. 1-way
   ANOVA, ∗p ≤ 0.05, ∗∗p ≤ 0.01 or ∗∗∗p ≤ 0.001.

3.4. Fibrotic CtsD^ΔMyel livers present enhanced inflammatory proteomic
signature

   Both Kupffer cells (KCs) and infiltrated bone marrow monocyte-derived
   macrophages (MoMϕs) play an essential role during the immune response
   followed by any liver injury, which directly contributes to the
   initiation and development of the fibrotic response. For that reason,
   we evaluated the inflammatory response during liver fibrosis in
   CtsD^F/F and CtsD^ΔMyel mice after both 8wk-CCl[4] administration and
   BDL. As expected, fibrotic livers displayed an increase in macrophages
   (F4/80+ve) ([139]Figures 5A and [140]S5A), despite no differences
   between CtsD^ΔMyel and CtsD^F/F mice were detected. Closer
   characterization of liver macrophages from fibrotic livers revealed
   that a significantly high proportion of macrophages (>80%) from
   8wk-CCl[4] livers were KCs (CLEC4F+ve-F4/80+ve) in both CtsD^F/F and
   CtsD^ΔMyel mice. Besides, no differences in this proportion were
   observed between phenotypes ([141]Figure 5B). Analysis of CtsD
   expression in CLEC4F+ve and CLEC4F-ve macrophages demonstrated that
   CtsD expression was significantly higher in KCs, which represented the
   main liver macrophage population in our model ([142]Figure 5C–D). As
   expected, almost no CtsD expression was detected in CtsD^ΔMyel liver
   macrophages ([143]Figures 5C–D and [144]S5B). Of note, fibrotic livers
   also displayed an increase in granulocytes (NIMP+ve) and neutrophils
   (MPO+ve) ([145]Fig. S5C–D), but no differences between phenotypes were
   detected. Similar results were obtained when analyzing neutrophil
   chemotactic proteins S100A8 and CXCL1 ([146]Fig. S5E–F). However,
   significantly higher levels of classical inflammatory mediators CCL2,
   CCL4 and TNF-α were displayed by 8wkCCl[4]-CtsD^ΔMyel livers in
   comparison with 8wkCCl[4]-CtsD^F/F ones ([147]Figure 5E). WB for CCL2
   in total liver also demonstrated an increase in 8wkCCl[4]-CtsD^ΔMyel
   livers despite not being statistically significant ([148]Fig. S5G).

Figure 5.

   [149]Figure 5
   [150]Open in a new tab

   CtsD^ΔMyelfibrotic livers present distinct and enriched immune and
   inflammatory proteomic profile.

   (A) Liver ADGRE1 (F4/80) gene expression in liver from 8-weeks Olive
   Oil (OOil) or CCl[4] treated CtsD^F/F and CtsD^ΔMyel mice and
   representative images of F4/80 IHC with zoomed area in liver tissue
   sections; (B) percentage CLEC4F+ve-F4/80+ve or CLEC4F-ve-F4/80+ve
   cells/field; (C) representative images of triple IF staining for F4/80
   (red), CLEC4F (cyan) and CtsD (green). White arrows point
   CLEC4F-ve-F4/80+ve cells; (D) CtsD MFI within CLEC4F+ve-F4/80+ve or
   CLEC4F-ve-F4/80+ve cells in liver tissue sections from 8-weeks CCl[4]
   treated CtsD^F/F and CtsD^ΔMyel mice. (E)CCL2, CCl4 and TNF-α gene
   expression in liver from 8-weeks Olive Oil (OOil) or CCl[4] treated
   CtsD^F/F and CtsD^ΔMyel mice. (F) Heatmap for the proteome distribution
   based on the Z-score-normalized LFQ abundances throughout all the
   samples. Green and red color ranges refer to lower and higher
   abundance, respectively; (G) Volcano plot of differentially regulated
   proteins. Green and red dots show the significant down- and
   up-regulated proteins, respectively, with highlight of CtsD protein;
   (H) Statistically significant enriched Reactome immune pathways in
   livers from CCl[4]-treated CtsD^ΔMyel versus CtsD^F/F mice; (I)
   Eulero-Venn diagram with common hits between the differential proteome
   of CtsD^ΔMyel mice and the genes of the immune system from the MGI
   database. Scale bars represent 50 μm. 1-way ANOVA, ∗p ≤ 0.05,
   ∗∗p ≤ 0.01 or ∗∗∗p ≤ 0.001.

   In agreement, early changes in gene expression were detected 7 days
   after BDL also demonstrating an increased expression of TNF-α, but not
   CCL2 and CCL4 ([151]Figs. S5H–J). A label-free quantitative (LFQ)
   proteomic analysis was performed comparing fibrotic
   8wkCCl[4]-CtsD^ΔMyel versus 8wkCCl[4]-CtsD^F/F livers, showing
   differential protein expression clustering for both phenotypes
   ([152]Figure 5F). The LFQ analysis quantified 1333 proteins (identified
   by a minimum of 2 unique peptides) and 189 proteins were detected as
   differentially expressed, of which 121 were up- and 68 down-regulated
   in the 8wkCCl[4]-CtsD^ΔMyel livers ([153]Figure 5G). Pathway enrichment
   analysis revealed significantly enriched Reactome immune pathways
   (p < 0.01) in fibrotic CtsD^ΔMyel livers ([154]Figure 5H). In addition,
   Eulero-Venn diagram intersecting the differential proteome of
   8wkCCl[4]-CtsD^ΔMyel livers and the genes of the immune system from the
   MGI database demonstrated 27 common hits, 22 of which were up-regulated
   ([155]Figure 5I). Of note, neutrophil degranulation pathway was also
   enriched in fibrotic CtsD^ΔMyel livers. Considering that CtsD deletion
   in CtsD^ΔMyel mouse might also affect neutrophils, we further
   investigated the degranulation capacity of neutrophils isolated from
   bone marrow of CtsD^ΔMyel mouse compared to CtsD^F/F ones. Activity
   analysis of secreted neutrophil elastase (NE), one of the main
   proteases stored in neutrophil azurophilic granules, after PMA
   stimulation showed no significant differences in neutrophil
   degranulation between phenotypes ([156]Fig. S5K).

   Our observations demonstrate an enhanced inflammatory signature in
   fibrotic CtsD^ΔMyel livers with no differences in the total numbers of
   liver macrophages or neutrophils, pointing towards a differential
   phenotypical change of CtsD^ΔMyel inflammatory cells as responsible for
   the enriched inflammatory proteomic profile.

3.5. CtsD deficient liver macrophages exhibit functional, phenotypical and
secretomic alterations compatible with impaired pro-resolutive shift

   Macrophages not only play an important role as drivers of the
   inflammatory response during liver fibrosis but can also participate in
   the remodeling and recycling of the extracellular matrix. Thus, we
   decided to analyze whether phenotypical changes in CtsD^ΔMyel
   macrophages also affected their collagenolytic capacity. First,
   Eulero-Venn analysis was performed combining the differential proteome
   of 8wkCCl[4]-CtsD^ΔMyel livers and the databases for the matrisome,
   cytoskeleton and lysosome, revealing 9, 22 and 5 common hits,
   respectively. Most of the hits identified were up-regulated, confirming
   an increased abundance of proteins within these signaling pathways in
   fibrotic CtsD^ΔMyel livers ([157]Fig. S6A). Next, collagenolytic
   activity of isolated liver macrophages was assessed using DQ Collagen I
   probe, which only emits fluorescence when is proteolytically cleaved.
   Proteolytic processing of DQ Collagen I was significantly decreased in
   CtsD^ΔMyel liver macrophages in comparison with CtsD^F/F ones
   ([158]Figure 6A) and partially colocalized with the lysosomal marker
   LAMP2, confirming DQ Collagen I processing was occurring within the
   lysosomes ([159]Figure 6B). To prove whether decreased DQ Collagen I
   signal was caused by impaired endocytosis, dextran internalization
   assay was performed in liver macrophages isolated from CtsD^F/F and
   CtsD^ΔMyel mice demonstrating similar levels of endocytosis independent
   of CtsD presence ([160]Figure 6C). In addition, no differences were
   observed in the expression of the collagen internalization receptor,
   UPARAP/Endo 180, its co-receptor UPAR and the lysosomal marker LAMP2 in
   8wkCCl[4]-CtsD^F/F and 8wkCCl[4]-CtsD^ΔMyel fibrotic livers
   ([161]Figure 6D). Supporting our results in liver macrophages, collagen
   I was detected within F4/80+ve macrophages in CCl[4]-treated livers
   ([162]Figure 6E) and CtsD showed a lysosomal distribution (LAMP-2) in
   liver macrophages ([163]Figure 6F). Pro-resolutive macrophages are
   responsible for ECM remodeling and recycling during fibrosis
   resolution. Hence, a decrease in the lysosomal collagenolytic activity
   of liver macrophages might be an indication of a wider defective
   pro-resolutive phenotype of macrophages. To study this, liver
   macrophages were isolated from CtsD^F/F and CtsD^ΔMyel CCl[4] treated
   mice. Indeed, CtsD^ΔMyel hepatic macrophages displayed a significant
   decrease in the expression of pro-resolutive markers such as CD206,
   TREM2 and TGF-β and an increase in the pro-inflammatory marker CD11c,
   with no changes in CD86 ([164]Figure 6G). Despite some reports indicate
   that CtsD is able to directly process collagen I [[165]19] and IV
   [[166]20], it is unlikely to be the only protease responsible for the
   phenotype observed in the CtsD^ΔMyel mice, as proteases do not operate
   alone but in networks of several different proteases. Thus, we decided
   to investigate CtsD^ΔMyel liver macrophage secretory protease signature
   to further elucidate CtsD associated protease network. Conditioned
   medias from hepatic macrophages isolated after an acute dose of CCl[4]
   were run in a protease array. Protease array detected 33 proteases, 18
   of which were differentially secreted in CtsD^ΔMyel liver macrophages
   ([167]Figures 6H–I and [168]S6B) and could be organized in an
   interactome ([169]Figs. S6C–D). Most of the proteases detected were
   down-regulated in CtsD^ΔMyel liver macrophages including MMP-2
   [[170]21,[171]22] and MMP-9 [[172][21], [173][22], [174][23]], which
   are both well-described gelatinases, responsible for the degradation of
   collagen during liver fibrosis. In addition, other less well-known
   enzymes with ECM degrading activity, such as CtsS [[175]24,[176]25],
   KLK7 [[177]26], MMP-3 [[178]21,[179]23] and MMP-7 [[180]23] were also
   significantly less secreted by CtsD^ΔMyel liver macrophages.
   Surprisingly, MMP-12 was the only enzyme whose secretion was
   significantly up-regulated in CtsD^ΔMyel hepatic macrophages.
   Interestingly, CtsD secreted by hepatic macrophages was proteolytically
   active and significantly downregulated in CtsD^ΔMyel mice
   ([181]Fig. S6E). Our results demonstrate CtsD^ΔMyel liver macrophages
   present deficient pro-resolutive characteristics, including defective
   lysosomal proteolytic processing of collagen, and altered secretomic
   proteolytic profile, which contributes to the enhanced liver fibrosis
   displayed by these mice.

Figure 6.

   [182]Figure 6
   [183]Open in a new tab

   CtsD^ΔMyelliver macrophages exhibit phenotypical and secretomic
   alterations compatible with impaired pro-resolutive shift.

   (A) Flow cytometry plot of relative cell counts and quantification of
   DQ collagen I expression in F4/80+ve liver macrophages isolated from
   CtsD^F/F and CtsD^ΔMyel mice. (B) Representative confocal images and
   colocalization mask of CtsD^F/F liver macrophages treated with DQCol I
   (green) and stained for LAMP-2 (red). (C) Flow cytometry plot of
   relative cell counts and quantification of Dextran internalization in
   F4/80+ve liver macrophages isolated from CtsD^F/F and CtsD^ΔMyel mice.
   (D) WB for Endo180, UPAR, LAMP2 and their respective β-actin in liver
   lysates from 8wkCCl[4]-CtsD^F/F and 8wkCCl[4]-CtsD^ΔMyel mice.
   Representative images of IF staining (E) for F4/80 (green) and Collagen
   I (red) and (F) for CtsD (green), LAMP-2 (red) and F4/80 (cyan) in
   liver tissue from 8wkCCl[4]-CtsD^F/F mice. (G)CD206, TREM2, TGF-β,
   CD11c and CD86 gene expression in liver macrophages isolated from
   CtsD^F/F and CtsD^ΔMyel mice treated with CCl[4] for 72 h. (H) Average
   protein expression level and (I) heatmap of proteases differentially
   regulated in conditioned media of KC isolated from CtsD^F/F or
   CtsD^ΔMyel mice treated 72 h with CCl[4]. Scale bars represent 10 μm
   for B and 20 μm for E-F. T-test analysis, ∗p ≤ 0.05, ∗∗p ≤ 0.01 or
   ∗∗∗p ≤ 0.001.

3.6. Changes in CtsD^ΔMyel liver macrophages result in impaired fibrosis
resolution

   To analyze whether these phenotypical and secretomic alterations in
   CtsD^ΔMyel liver macrophages translated into changes in fibrosis
   resolution in vivo, we administered CCl[4] for 4 weeks and established
   two endpoints after the last dose of CCl[4], 1 day for the peak-CCl[4]
   and 3 days for the resolution-CCl[4] ([184]Figure 7A). The percentage
   of liver collagen remodeling during resolution was assessed by
   hydroxyproline ([185]Figure 7B) and liver staining of denatured
   collagen chains using fluorescently labelled Gly-X-Y collagen
   hybridizing peptides ([186]Figure 7C) demonstrating that collagen
   remodeling and degradation was significantly decreased during
   resolution in CtsD^ΔMyel mice. Further study of liver tissues during
   the resolution phase showed that, despite no significant differences
   were observed in the total area occupied by F4/80+ve cells, the
   percentage of macrophages surrounding the fibrotic area was decreased
   in the CtsD^ΔMyel livers, supporting the idea that macrophages are not
   properly fulfilling their role during resolution ([187]Figure 7D). In
   addition, and opposite to our observation in the 8wk-CCl[4] livers,
   during the resolution phase CtsD expression was significantly higher in
   MoMϕs (CLECF4-ve-F4/80+ve) instead of in the KCs ([188]Figure 7E–F),
   despite similar proportion of KCs and MoMϕs were maintained as in the
   8wk-CCl[4] model ([189]Figure 7G). Closer analysis of the lysosomal
   content of liver macrophages during the resolution phase demonstrated
   similar levels between phenotypes ([190]Figure 7H–I), but analysis by
   TEM revealed the presence of abnormal lysosomal structures in
   CtsD^ΔMyel liver macrophages from normal adult livers ([191]Figure 7J).
   These findings might underpin an alteration on lysosomal function, but
   not lysosome numbers, as responsible for the secretomic and
   phenotypical changes observed in CtsD^ΔMyel mice. Besides the direct
   effect that CtsD^ΔMyel liver macrophage altered secretome
   ([192]Figures 6H–I and [193]S6C,D) can have over ECM remodeling, the
   imbalance in extracellular protease content might also exert an
   indirect effect over the function of proteins that participate in the
   ECM assembly and organization. SPARC (secreted protein acidic and rich
   in cysteine), which is a collagen-binding matricellular protein that
   participates in collagen I assembly, exhibits 13 cleaving sites for
   CtsD, spanning 18–303 residue range ([194]Table S3) [[195]27]. Analysis
   of SPARC expression within the collagen I area demonstrated that while
   in CtsD^F/F mice there was a significant reduction in the expression of
   SPARC associated with collagen during resolution, no changes were
   observed in CtsD^ΔMyel mice ([196]Figure 7K). These results might
   indicate that during fibrosis resolution active CtsD secreted by
   macrophages ([197]Fig. S6E) could proteolytically process SPARC,
   facilitating ECM remodeling and degradation ([198]Figure 7K). Lack of
   secreted active CtsD would result in sustained SPARC-collagen
   association, contributing to fibrosis perpetuation. Our results
   revealed that phenotypical and secretomic alterations of CtsD^ΔMyel
   macrophages caused suboptimal collagen remodeling in vivo, hampering
   fibrosis resolution and perpetuating liver fibrosis.

Figure 7.

   [199]Figure 7
   [200]Open in a new tab

   Phenotypical and secretomic alterations of CtsD^ΔMyelliver macrophages
   causes deficient fibrosis resolution in vivo.

   (A) Diagram displaying experimental setting for the fibrosis resolution
   model. (B) Percentage of HP collagen remodeling in liver tissue, (C)
   percentage of mean fluorescent intensity (MFI) of R-CHP staining in
   liver sections and representative R-CHP images from resolution-CCl[4]
   CtsD^F/F and CtsD^ΔMyel mice. (D) Percentage F4/80 area in total area
   or fibrotic area, (E) representative images of triple IF staining for
   F4/80 (red), CLEC4F (cyan) and CtsD (green). White arrows point
   CLEC4F-ve-F4/80+ve cells; (F) CtsD MFI within CLEC4F+ve-F4/80+ve or
   CLEC4F-ve-F4/80+ve cells; (G) percentage CLEC4F+ve-F4/80+ve or
   CLEC4F-ve-F4/80+ve cells/field in liver tissue sections from
   resolution-CCl[4] CtsD^F/F and CtsD^ΔMyel mice. (H) Representative
   images of dual IF of F4/80 (green) and LAMP2(R) with zoomed area and
   (I) LAMP2 MFI within F4/80 area in liver sections from
   resolution-CCl[4]CtsD^F/F and CtsD^ΔMyel mice. (J) Representative liver
   images of transmission electron microscopy displaying liver macrophages
   with zoom in lysosomal structures from CtsD^F/F and CtsD^ΔMyel mice.
   (K) SPARC MFI within R-CHP area in liver sections from peak- and
   resolution-CCl[4] CtsD^F/F and CtsD^ΔMyel mice and representative
   images of dual IF staining for R-CHP (red) and SPARC (green) in liver
   sections from peak- and resolution-CCl[4] CtsD^F/F mice. Scale bars
   represent 50 μm. T-test analysis, ∗p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗p ≤ 0.001.

3.7. Human cirrhotic CtsD-expressing MPs subpopulations are enriched in ECM
degradation pathways

   To investigate the potential biological translation of our results in
   mouse liver fibrosis into human cirrhosis, we performed further
   analysis of the single-cell transcriptomes [[201]14]. Specific analysis
   of CtsD-expressing MPs identified KCs and SAM(2) as highly expressing
   CtsD ([202]Figure 8A and [203]Fig. S7A–G) in human cirrhotic livers.
   Analysis of differentially expressed transcripts between
   CtsD-expressing and non-expressing MPs in cirrhotic livers revealed 17
   differentially expressed genes, 16 of which were upregulated and
   included the previously described pro-resolutive marker GPNMB
   [[204]28,[205]29] ([206]Figure 8B). Additional heatmap analysis
   demonstrated a distinct gene expression pattern of the highest-CtsD
   expressing MPs subpopulations (KC(1), KC(2) and SAM(2)) in cirrhotic
   livers ([207]Fig. S8A). Volcano plot comparing the differential
   expression of these three MPs subpopulations versus the other MPs
   confirmed these subpopulations as the ones highly expressing the
   macrophage pro-resolutive markers FABP5, GPNMB and TREM-2
   [[208]28,[209]29] ([210]Fig. S8B). Further analysis of differentially
   regulated pathways in CtsD-expressing versus non-expressing MPs
   revealed an enrichment in genes related to the ECM content and
   degradation and the lysosome ([211]Figure 8C–D). In agreement with
   these results, gene set enrichment analysis of the transcriptome of
   human cirrhotic KCs [[212]16] confirmed an enrichment in pathways
   related to the lytic organization, the endopeptidase complex and the
   lysosomal lumen ([213]Figure 8E). Our results demonstrate that
   CtsD-expressing MPs in human cirrhosis display an enriched gene
   expression signature in the lysosome and the ECM degradation related
   pathways.

Figure 8.

   [214]Figure 8
   [215]Open in a new tab

   CtsD-expressing MPs in cirrhotic patients are associated to ECM
   degradation in human liver cirrhosis.

   (A) UMAP projection depicting CtsD expression and dotplot representing
   percentage CtsD expression within the different MPs subpopulations in
   human cirrhotic livers ([216]GSE136103). (B) Volcano plots for
   differential expression of CtsD-expressing MP versus
   CtsD-non-expressing MPs in cirrhotic livers. (C) Top 10 up and down
   regulated signalling pathways and (D) selected pathways in
   CtsD-expressing MPs versus CtsD-non-expressing MPs in cirrhotic livers.
   (E) Selected gene set enrichment analysis within the 15 pathways with
   highest NES of human cirrhotic KCs ([217]GSE123661).

4. Discussion

   In this study we demonstrate that CtsD-driven lysosomal activity in
   macrophages is essential for liver fibrosis development and resolution.
   Liver fibrosis is caused by a dysregulation of the normal homeostatic
   wound healing response due to chronic injury and can lead to severe
   organ dysfunction or even failure, as a result of the replacement of
   functional parenchyma by fibrotic ECM. Liver fibrosis is currently
   widely accepted to be reversible, however, no specific treatments to
   slow-down or reverse fibrosis have been successfully developed so far.
   Little is still known about the biological mechanisms controlling ECM
   remodeling during liver fibrosis. Changes in proteolytic activity are
   essential to liver fibrosis development and resolution as ECM
   homeostasis is controlled, among others, by proteases. The contribution
   of cysteine cathepsins to the development of fibrotic diseases has been
   previously reported in heart (CtsL) [[218]30], lung (CtsK) [[219]31],
   and liver (CtsB) [[220]3]. However, the role of aspartyl cathepsins,
   such as CtsD, has not yet been explored.

   Analysis of CtsD expression in control and cirrhotic human scRNAseq
   datasets [[221]14] and liver biopsies revealed CtsD expression in
   hepatocytes and macrophages ([222]Figure 1A–C). Analysis of CtsD in KCs
   (Log2 Fc = 2.63 p ≤ 0.001) and in serum ([223]Figure 1D) from cirrhotic
   patients revealed a significant increase of CtsD during cirrhosis. It
   is not surprising that CtsD major reservoir in the liver are
   macrophages, as they have been reported to display increased
   proteolytic activity during liver injury [[224]32] and proteolytic
   enzymes are essential for processing phagocytic products and for
   macrophage mesenchymal migration [[225]33]. To investigate the
   cell-specific role of CtsD in macrophages or in hepatocytes during
   liver fibrosis we generated two novel cell-specific KO mouse strains by
   breeding CtsD floxed [[226]9] mice with LysMCre mice [[227]10]
   (CtsD^ΔMyel) or AlbCre [[228]11] (CtsD^ΔHep) mice and validated CtsD
   deletion in liver macrophages and hepatocytes, respectively
   ([229]Figures 2 and [230]S2). Liver fibrosis was induced by chronic
   CCl[4] administration or BDL. Enhanced liver fibrosis was demonstrated
   by an increase in collagen deposition ([231]Figure 3C–D, [232]Fig. S3F
   and [233]Figure 4C–D) and fibrogenic markers ([234]Figure 3E–H, S3G-H
   and S4E-G) in liver tissue from CtsD^ΔMyel mice, but not in CtsD^ΔHep
   ones ([235]Figures 3L-N, 4K-M, S3A-B and S4A-B), after both CCl[4] and
   BDL induced liver fibrosis without affectation of the liver damage
   ([236]Figurs 3I, [237]S3C, [238]4H, [239]S4C and S4H). When we assessed
   liver inflammation, we observed an increase in gene expression of
   inflammatory markers (CCL2, CCL4 and TNF-α) ([240]Figures 5E and S5H-J)
   and a distinct proteomic signature ([241]Figure 5F–G) with a
   significant enrichment in the Reactome immune pathways in
   8wkCCl[4]-CtsD^ΔMyel fibrotic livers ([242]Figure 5H–I). No significant
   differences were observed in the number of macrophages ([243]Figures 5A
   and [244]S5A) and the proportion of KCs and MoMϕs ([245]Figure 5B)
   detected in fibrotic liver tissue between CtsD^F/F and CtsD^ΔMyel mice.
   Interestingly, CtsD expression was significantly higher in KCs
   (CLECF4+ve-F4/80+ve) ([246]Figure 5C–D), which represented the main
   liver macrophage population in our model. These results suggest that
   functional or phenotypical changes of the inflammatory cell population
   might be responsible for the enriched inflammatory signature displayed
   in the liver in the absence of CtsD, instead of changes in the number
   or proportion of liver macrophages. While the role of macrophages
   during initiation, progression and resolution of liver fibrosis is
   quite well defined, the role of neutrophils is still controversial
   [[247]34]. Regarding our work, it is important to notice Cre
   recombination driven by lysozyme 2 gene promoter also occurs in
   granulocytes, including neutrophils. Our results demonstrated that
   neither the number of granulocytes or neutrophils ([248]Figs. S5C–D),
   nor the expression of neutrophil chemotactic proteins
   ([249]Figs. S5E–F), or their degranulation capacity ([250]Fig. S5K) are
   different between phenotypes. Nevertheless, although several reports
   have shown that neutrophil depletion during chronic liver disease does
   not influence liver fibrosis [[251]35], we still cannot discard that
   CtsD deletion in neutrophils might also partially contribute to the
   phenotype observed in CtsD^ΔMyel mice. Thus, further studies will be
   needed to clarify the full impact of CtsD deletion in neutrophils over
   liver fibrosis, as this was not the aim of our study.

   During liver fibrosis, macrophage plasticity allows them to adapt their
   phenotype, playing different and even opposing roles in disease
   progression. Hence, macrophages are responsible for the immune and
   inflammatory mechanisms driving liver fibrosis but also promote tissue
   repair and remodeling, contributing to fibrosis resolution. To analyze
   the effect of CtsD deletion in macrophages over ECM remodeling, we
   intersected the 8wkCCl[4]-CtsD^ΔMyel differential proteome with the
   matrisome database, confirming increased abundance of ECM-related
   proteins in fibrotic CtsD^ΔMyel livers ([252]Fig. S6A). Little is still
   known about the function and phenotypical characteristics of
   restorative macrophages. While their indirect role modulating ECM
   remodeling due to their ability to secrete proteolytic enzymes, such as
   MMPs, is been recognized [[253]28], their role as fibrolytic cells,
   directly phagocytosing and degrading ECM proteins, is far less
   understood [[254]36]. Our experiments demonstrated that CtsD is
   important for lysosomal proteolytic processing of DQ Collagen I in
   liver macrophages ([255]Figure 6A–B). Impaired lysosomal proteolysis
   has also been reported due to the absence of CtsE in macrophages
   [[256]37] supporting our findings. Receptor-mediated collagen
   endocytosis is mediated in macrophages by the uPAR-associated protein
   (uPARAP/Endo180) in cooperation with UPAR [[257]38,[258]39]. No
   differences in endocytosis in hepatic macrophages ([259]Figure 6C) or
   the expression of collagen internalization receptor UPARAP/Endo180, and
   its partner UPAR ([260]Figure 6D) were detected in CtsD^F/F and
   CtsD^ΔMyel fibrotic livers, suggesting that changes in internalized DQ
   Collagen I signal were caused by defective proteolytic processing
   instead of alterations in the receptor-mediated endocytic pathway.
   Indeed, collagen I was visualized within macrophages in fibrotic livers
   demonstrating collagen I was being internalized by liver macrophages
   in vivo during liver fibrosis ([261]Figure 6E). Furthermore,
   confirmation of CtsD lysosomal distribution in liver macrophages
   ([262]Figure 6F) from fibrotic livers supported an intracellular role
   for CtsD in collagen remodeling.

   Since pro-resolutive macrophages are responsible for ECM remodeling and
   recycling during fibrosis resolution, a decrease in the lysosomal
   collagenolytic activity of liver macrophages might be an indication of
   a wider defective pro-resolutive phenotype of macrophages. Analysis of
   CtsD^ΔMyel hepatic macrophages demonstrated decreased pro-resolutive
   markers such as CD206, TREM2 [[263]29] and TGF-β and an increase in the
   pro-inflammatory marker CD11c ([264]Figure 6G). In line with our
   findings, strong enrichment of CtsD has been previously reported in
   TREM+ve macrophage subclusters during murine NASH [[265]29], displaying
   pro-resolutive markers resembling those previously described in
   pre-monocyte-derived KCs and MMP12+ve macrophages [[266]29,[267]40]. In
   agreement with our findings, combined pharmacological inhibition of
   cathepsins B, L and S in tumor associated macrophages from pancreatic
   cancer resulted in phenotypical shifting from M2 to M1 linked to
   distinct lysosomal alterations, however, the exact mechanism is still
   unknown [[268]41].

   Despite direct proteolytic processing of collagen I^19 and IV [[269]20]
   by CtsD has been previously reported, it is unlikely to be the only
   protease responsible for the defective collagen remodeling observed in
   CtsD^ΔMyel fibrotic livers. Indeed, protease array from hepatic
   macrophage conditioned medias revealed 18 out of 33 proteases
   differentially regulated in CtsD^ΔMyel liver macrophages, most of them
   downregulated, which could be built in an interactome providing novel
   insights on hepatic macrophage protease network ([270]Figure 6H–I and
   S6C-D). Among the proteases with decreased secretion were
   well-described gelatinases MMP-2 [[271]21,[272]22] and MMP-9
   [[273][21], [274][22], [275][23]] and less well-known enzymes with ECM
   degrading activity, such as CtsS [[276]24,[277]25], KLK7 [[278]26],
   MMP-3 [[279]21,[280]23] and MMP-7 [[281]23]. Interestingly, MMP-12 was
   the only enzyme whose secretion was significantly up-regulated in
   CtsD^ΔMyel liver macrophages, opposite to what could be expected from
   recent reports proposing MMP-12 as a maker for restorative macrophages
   in murine experimental models of liver fibrosis [[282]28,[283]29].
   Since MMP-12 is synthetized as an inactive pro-form (zymogen), gene
   transcription is not the first regulatory step, but the proteolytic
   activation of the existing cellular pool. Hence, opposite observations
   at the transcript, protein and even activity level are not unusual when
   it comes to protease study, as changes in gene transcription not always
   encompass the same changes at the protein or activity level. Thus,
   further study is required to fully clarify the role of MMP-12 in
   restorative macrophages not only as a marker but also as part of their
   proteolytic landscape. Phenotypical and secretomic alterations of
   CtsD^ΔMyel liver macrophages resulted in deficient fibrosis resolution
   in vivo ([284]Figure 7A–C) and decreased number of macrophages in the
   fibrotic areas ([285]Figure 7D) supporting macrophages might not be
   fulfilling their role during resolution. Interestingly, a switch in
   CtsD expression was observed from KCs (CLECF4+ve-F4/80+ve)
   ([286]Figure 5C–D) to MoMϕs (CLECF4-ve-F4/80+ve) ([287]Figure 7E–F)
   during the resolution phase, despite similar proportion of KCs and
   MoMϕs ([288]Figure 5, [289]Figure 7G) being maintained. Despite no
   change in lysosomal content ([290]Figure 7H–I) of macrophages was
   observed during liver fibrosis resolution, TEM analysis of normal mouse
   livers revealed already the presence of abnormal lysosomal structures
   in CtsD^ΔMyel liver macrophages ([291]Figure 7J), underpinning an
   alteration on lysosomal function, but not lysosomal content, as
   responsible for the secretomic and phenotypical changes observed in
   CtsD^ΔMyel mice. Similarly, mice with lysosomal dysfunction due to the
   loss of glycosylated lysosomal membrane protein (GLMP) displayed
   spontaneous [[292]42] and aged-dependent [[293]43] liver fibrosis,
   pointing out correct lysosomal activity as an essential mechanism of
   control for liver fibrosis resolution. Complementary to its role within
   the lysosome, our experiments also demonstrated that CtsD can be
   actively secreted by liver macrophages ([294]Fig. S6E). As an active
   secreted enzyme, CtsD can proteolytically process not only ECM
   proteins, but also proteins involved in ECM assembly and stabilization
   such as SPARC, which is one of CtsD substrates exhibiting 13 cleaving
   sites ([295]Table S3). Considering its important role in fibril
   aggregation and growth, it is not surprising that SPARC knock-out mouse
   displays reduced collagen deposition and fibrosis in multiple organs
   including the liver [[296]44]. Our experiments demonstrate that the
   amount of SPARC located within collagen I was significantly decreased
   during the resolution phase only in CtsD^F/F mouse ([297]Figure 7K).
   This result suggests that the lack of active secreted CtsD could be
   responsible for sustained SPARC-collagen association, which would also
   contribute to fibrosis perpetuation.

   Analysis of human scRNAseq cirrhotic datasets identified KC(1), KC(2)
   and SAM(2) as the MPs subpopulations with the highest CtsD expression
   ([298]Figure 8A), and with a distinct gene expression pattern compared
   to the other MPs subpopulations in cirrhotic livers ([299]Figures 8B
   and [300]S8A). Interestingly, FABP5, GPNMB and TREM-2, which have been
   proposed as markers of restorative macrophages [[301]28,[302]29], were
   up-regulated in these three MPs subpopulations ([303]Fig. S8B).
   Analysis of differentially regulated pathways in CtsD-expressing versus
   non-expressing MPs revealed an enrichment in genes related to both the
   ECM content and degradation and the lysosome ([304]Figure 8C–D). This
   was confirmed by gene set enrichment analysis of human cirrhotic KC
   transcriptome [[305]16] ([306]Figure 8E) in agreement with previous
   reports, restorative macrophages present enriched signature in
   lysosomal signaling pathways [[307]28]. Thus, our results demonstrate
   that CtsD-expressing MPs in human cirrhosis display an enriched gene
   expression signature in the lysosome and the ECM degradation related
   pathways.

   Our work provides the first description of how lysosomal function
   driven by CtsD modulates macrophage resolutive capacity and
   fibronolytic activity. Despite further work needs to be done to fully
   elucidate the multiple roles of the macrophage proteolytic network
   controlling ECM remodeling, our results put in the spotlight for the
   first time the importance of the lysosomal protease balance for the
   correct functioning of macrophages during fibrosis resolution. This
   knowledge is the first critical step to start exploring, designing and
   testing novel drugs based on proteases that will enhance ECM
   degradation and promote fibrosis resolution. In summary, our work
   demonstrates CtsD-driven lysosomal activity is essential for the
   degradomic shift, which comprises all the secretomic, phenotypical and
   functional changes required by macrophages to proteolytically process
   the ECM during liver fibrosis resolution. Our findings start to
   elucidate a novel and unknown degradome landscape for restorative
   macrophages in liver disease.

Financial support

   This work was mainly funded by MCIN/AEI/10.13039/501100011033/FEDER, UE
   through the project grants PID2021-123652OB-I00 and
   RTI2018-097475-A-100 (AM); by MCIN/AEI/10.13039/501100011033 and El FSE
   invest in your future through the contract RYC-2016-19731 (AM); by
   Pfizer grant #77131383 (AM); by Ministerio de Universidades fellowships
   FPU19/05357 (PRB) and FPU20/01367 (MFF); by
   MCIN/AEI/10.13039/501100011033/FEDER, UE contract PRE2022-101676 (JCP)
   and CSIC contract JAEINT_23_01245 (AdCC). Additional funding sources
   can be found in the financial support and sponsorship section of the
   manuscript.

CRediT authorship contribution statement

   Paloma Ruiz-Blázquez: Writing – review & editing, Methodology,
   Investigation, Formal analysis. María Fernández-Fernández: Writing –
   review & editing, Methodology, Investigation, Formal analysis. Valeria
   Pistorio: Methodology, Investigation, Formal analysis. Celia
   Martinez-Sanchez: Methodology, Investigation. Michele Costanzo: Writing
   – review & editing, Methodology, Investigation. Paula Iruzubieta:
   Writing – review & editing, Resources, Methodology. Ekaterina
   Zhuravleva: Methodology, Investigation. Júlia Cacho-Pujol: Methodology,
   Investigation. Silvia Ariño: Investigation, Methodology. Alejandro Del
   Castillo-Cruz: Investigation, Methodology. Susana Núñez: Methodology.
   Jesper B. Andersen: Resources. Margherita Ruoppolo: Writing – review &
   editing, Resources, Methodology. Javier Crespo: Resources. Carmen
   García-Ruiz: Resources. Luigi Michele Pavone: Writing – review &
   editing, Resources, Methodology. Thomas Reinheckel: Resources. Pau
   Sancho-Bru: Resources. Mar Coll: Resources, Methodology. José C.
   Fernández-Checa: Resources. Anna Moles: Writing – review & editing,
   Writing – original draft, Visualization, Validation, Supervision,
   Software, Resources, Project administration, Methodology,
   Investigation, Funding acquisition, Formal analysis, Data curation,
   Conceptualization.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgments