**4. Discussion**

In light of recent translational approaches in the field of liver fibrosis, a thorough understanding of fibrocyte biology is urgently needed. Bone marrow transplantation, for instance, proved beneficial in murine cholestatic fibrosis but remains controversial as a treatment option for chronic liver diseases [48]; Cenicriviroc (CVC), a dual CCR2/CCR5-inhibitor impeding the infiltration of monocytes, is a promising drug candidate for NAFLD patients with fibrosis [49,50]. Fibrocytes potentially contribute to the targeted monocyte population and are known to be recruited via CCR2, -3, -5, and -7 signaling [8,20,23,51,52]. CVC just recently provided promising results reducing fibrosis but surprisingly not reducing inflammation in NAFLD-patients, and is currently tested in a phase III trial (NCT03028740) [53].

We herein present a novel approach to investigate the role of bone marrow-derived fibrocytes in liver fibrosis. Our model enabled the specific depletion of fibrocytes, avoiding the dependence on particular surface markers or differentiation pathways. RNA in situ hybridization, however, implies that it did not entirely deplete bone marrow-derived fibrocytes (Figure 1b). This result appears in line with evidence from a study Puche et al. conducted: utilizing the HSV-TK model, about 65% of HSCs could be depleted in a CCl4-model of hepatic fibrosis [54]. Even though our analyses suggest a superior depletion-rate, it should be a concern of future studies to closely monitor the depletion effectiveness to not underestimate the role of bone marrow-derived fibrocytes.

Our results show a functional contribution of fibrocytes to hepatic fibrogenesis. The determination of hydroxyproline content revealed a reduced deposition of fibrillar collagens as a result of the depletion of fibrocytes (Figure 2c). The present data allow a range of interpretations regarding the extent of the mitigation: While the common prediction, mainly based on fate-tracing studies and evidence from other organs, that the depletion of fibrocytes yields minor effects on hepatic fibrogenesis [15,17,55] is compatible with our data, the 95% confidence interval also spans a reduction of up to ∼15%. An attenuation of that magnitude is considered highly clinically significant and would challenge our understanding of the contribution of fibrocytes. Significantly reduced serum ALT levels (Figure 5a), despite the generally moderate level of hepatocyte damage, support this notion. It appears noteworthy that (1) the reduction of hydroxyproline was not accompanied by a changed gene expression of collagen I in our study and (2) contradicting results were obtained in studies, investigating the contribution of fibrocytes to fibrosis of the liver [39] and lung [56].

The unchanged gene expression of *Col1a1*, *Col1a2,* and *Col3a1* (Figures 2e and S3) implies that a reduced secretion of collagens at the time of analysis is not the cause of the reduced hydroxyproline content. The development of fibrosis is highly dependent upon the balance of deposition and degradation of ECM-material [57]. Although fibrocytes are known to express several MMPs [30,31], the overall hepatic expression of these was unchanged in result of the fibrocyte ablation (Figure 2f), too. We therefore hypothesize that the mitigation of fibrosis is the result of a transient regulation of fibrogenesis during the disease progression. Given their properties as hematopoietic, circulating cells, fibrocytes can be found early at the site of injury [3,20]. Investigations regarding the influx of fibrocytes into the injured liver show a peak two weeks after onset of the fibrogenic stimulus [20]. It has to be considered, however, that fibrosis was induced via CCl4 in this experiment, which provokes an accelerated disease progression, compared to the TAA model [41,42]. Taken together, these results suggest that our timepoint of analysis missed the greatest contribution of fibrocytes yet displayed a lasting effect and provide encouragement to more closely focus on the role of fibrocytes in different stages of disease progression in future research.

Ozono et al. just recently published their findings as a clodronate liposome-mediated depletion of fibrocytes »had little contribution on liver fibrosis« in a murine model of CCl4-induced fibrosis [39]. Although the authors concluded differently, we argue that the results they present are not necessarily contradictory to ours. Morphometric analysis displays a reduction of stained fibrillar collagens by tendency (see Figure 3b in reference [39]). Smaller sample sizes (*n* = 8) might provide an explanation of why the level of statistical significance postulated by the authors was not reached. Furthermore, semiquantitative means like histology with subsequent pathological evaluation (staging) or morphometric analysis perhaps lack the accuracy to detect subtle changes in the deposition of ECM components. Consistent with this claim, semiquantitative methods failed to detect the mitigation of fibrosis in our study (Table 1, Figure 2d). Solely relying on those techniques might under some circumstances therefore be inadequate to elaborate the biology of fibrocytes. Even though it will be inevitable to study the contribution of fibrocytes in different models of fibrosis, the use of distinct models and readout parameters impedes the comparability of results obtained with such.

A specific knockout of the *Col1a1* gene in fibrocytes, furthermore, yielded no significant impact on pulmonary fibrosis, even though up to 30% of collagen producing cells are assumed to be fibrocytes, suggesting a crucial role of paracrine functions [56]. We herein sought to investigate effects on the activation and proliferation of myofibroblasts, hepatic inflammation, and cell death. Since there is compelling evidence for activated HSCs being the main contributors to hepatic collagen deposition [15–17], and fibrocytes, in fact, can facilitate the activation of myofibroblast via the secretion of TGF-β and PDGF in vitro [26], a decreased activation of myofibroblasts might provide a plausible explanation for the observed attenuation of fibrosis. Even though a transient process cannot be excluded, our results, showing an unchanged expression of α-SMA, *Tgfb,* and *Pdgfb* in result of the fibrocyte depletion (Figure 3), tend to refute this hypothesis. The reduced CD45-stained area (Figure 4a,b) and the decreased hepatic concentration of IL-1β (Figure 4c) might imply an ameliorated inflammatory response in consequence of the fibrocyte depletion. Nevertheless, the evaluation of inflammatory cytokines (Figures 4c,d and S7) emphasized that bone marrow-derived fibrocytes are not a major source of inflammatory mediators at the time of analysis. These results are noteworthy, given the fact that previous research provided evidence for a participation of fibrocytes in inflammatory processes [26,34,58] and entities like scleroderma, rheumatoid diseases, and asthma are associated with fibrocytes (reviewed in reference [24]). Our findings call for careful considerations, especially regarding the interpretation of cultivation and stimulation experiments performed with fibrocytes. Lastly, the decreased levels of serum ALT (Figure 5a) can be interpreted as a result of ameliorated fibrosis. They might, however, also display an attenuation of hepatic cell death, caused by the depletion of fibrocytes, leading to reduced profibrogenic stimuli. Serum amyloid P, which is known to inhibit the differentiation of fibrocytes [59], prevented hepatic cell damage in CCl4-induced acute liver injury [60]. While multiple forms of hepatic cell death are known [61], we found subtle regulations regarding apoptosis. Contradictory and partly not reproducible results herein forbid a conclusive interpretation.

In summary, we herein demonstrate a functional contribution of bone marrow-derived fibrocytes to hepatic fibrogenesis. However, a definitive mode of action could not be identified. It has to be considered that neither our analyses of paracrine fibrocyte functions, despite covering the crucial mediators of hepatic fibrogenesis, were exhaustive nor the previous cultivation and fate-tracing studies necessarily elucidated the entire range of fibrocyte functions in a complex in vivo setting. Since it is, due to the high plasticity and little number of cells, often troublesome to study fibrocytes in vivo, it is noteworthy that properly planned animal experiments with a rigorous statistically substantiated design according to the 3R principles enabled the generation of robust results. Fibrocytes should be considered in future research to acquire a thorough understanding of the biology of hepatic fibrosis.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/10/1210/s1, Figure S1: Schematic gene construction and basic experimental data, Table S2: Primer sequences used in quantitative real-time PCR, Figure S3: Gene Expression Array results, Figure S4: Hepatic protein concentrations of MMPs and TIMPs, Figure S5: Detailed quantitative real-time PCR results, Table S6: Grading according to Ishak et al., Figure S7: Proteome Profiling of inflammatory cytokines, Figure S8: Hepatic protein concentrations of inflammatory cytokines.

**Author Contributions:** Conceptualization, M.R. and E.R.; Data curation, F.H. and M.R.; Formal analysis, F.H., M.R. and J.P.-K.; Funding acquisition, M.R.; Investigation, F.H., M.R., A.S., K.I., K.K., Y.C. and J.B.; Methodology, F.H., M.R., R.S. (Rajkumar Savai), R.V., L.K. and J.B.; Project administration, M.R. and E.R.; Resources, M.R., R.S. (Ralph Schermuly), K.K. and E.R.; Supervision, M.R.; Validation, F.H.; Visualization, F.H.; Writing—original draft, F.H. and M.R.; Writing—review & editing, F.H., M.R., K.I., L.K., J.B. and E.R.

**Funding:** This work was supported by grants from the German Research Foundation (RO 957/10-1 to E.R.), Max Planck Society, Cardio-Pulmonary Institute (CPI), and the German Center for Lung Research (DZL). F.H. received starting grants from the Justus Liebig University Giessen.

**Acknowledgments:** The authors thank Dirk Krambrich for performing irradiation experiments and Annette Tschuschner, Heike Müller, and Dagmar Leder for excellent technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.
