Tissue LyC6⁻ Macrophages Are Generated in the Absence of Circulating LyC6⁻ Monocytes and Nur77 in a Model of Muscle Regeneration

Monocytes in blood form a heterogeneous group of cells that are precursors of a number of tissue-resident cells, including macrophages (MFs) and certain dendritic cell (DC) subtypes (1, 2). Blood monocytes in both humans and mice can be grouped into two major and distinct subtypes (3). The discrimination of monocyte subsets is based on their anatomical location and differential expression of cell surface markers, with the most typical markers used for discrimination in the mouse being Ly6C, chemokine receptors such as CCR2 and CX3CR1, and the adhesion molecule SELL (CD62L). Importantly, the two major subsets distinguished by cell surface markers do not only represent minor phenotypic variation of a canonical blood monocyte population; rather, it is believed that these two distinct monocyte populations differ in their origin, differentiation, cell fate, and immune functions (4, 5). This is highlighted by the distinct roles of Ly6C⁺ and LyC6⁻ monocytes in a number of pathological conditions such as microbial infections (6), atherosclerosis (7, 8), and myocardial healing (9). Recently, characterization of monocyte subsets by cellular function and/or transcriptional regulators has gained momentum and holds the potential for a better understanding of monocyte and MF development (10–12).

Similarly to blood monocytes, tissue MFs, which are generated from precursor and/or monocyte populations, can be divided into Ly6C⁺ and Ly6C⁻ subsets. The origin, cell fate, and functions of distinct MF subsets, generated either under steady-state or inflammatory conditions, are believed to be characteristically different. At least a subset of steady-state F4/80^bright MFs appears to derive from the yolk sac (13) and is differentiated along a distinct pathway compared with MF subsets derived from hematopoietic stem cells. Relatively little definitive information is available on the ancestry of other resting or inflammatory Ly6C⁺ and Ly6C⁻ MF subsets.

Several key issues about monocyte and MF development and the differential development of Ly6C⁺ and LyC6⁻ cells are not well understood and/or debated. One of the main questions pertains to the developmental relationship between bone marrow precursors and peripheral Ly6C⁺ and LyC6⁻ monocytes and/or MFs. The second, related question is what is the developmental relationship between blood and tissue Ly6C⁺ and LyC6⁻ cells? In principle, when discussing the first question, a few alternative explanations can be provided for the existence of the two monocyte subsets. One possible, but experimentally not yet well-supported explanation is that Ly6C⁺ and LyC6⁻ monocytes are derived from clearly separate lineages. Another theoretical explanation presumes the existence of a common bone marrow precursor that gives rise to both monocyte subsets independently. Such common precursors that can give rise to different MF and monocyte subsets have been found (14), but their contribution to LyC6⁻ lineages has not been clarified. A third, partly overlapping explanation for the generation of the two monocyte subsets states that circulating Ly6C⁺ monocytes serve as the precursors for LyC6⁻ monocytes. An emerging body of experimental data supports this latter scenario (15). An analogous question is left unanswered in the case of tissue MFs, as well. It is incompletely understood which precursor or blood monocyte populations give rise to tissue Ly6C⁺ and Ly6C⁻ MFs and whether tissue Ly6C⁺ MFs can serve as local precursor for the development of Ly6C⁻ MFs. In addition, the question could be asked whether the developmental history and relationship of blood Ly6C⁺ and LyC6⁻ monocytes are mirrored in the development of tissue Ly6C⁺ and Ly6C⁻ MFs.

Finding answers to these questions is hampered by the lack of subset-specific markers that could allow one to trace precursor populations and Ly6C⁺ and LyC6⁻ subsets during their development. However, the identification of transcription factors required for the generation of certain subtypes can aid this work and can be used (i.e., as genetic means) to dissect developmental relationships. A recent report elegantly demonstrated (16) that the LyC6⁻ blood monocyte compartment is severely diminished in full-body Nr4a1 (NUR77) knockout (KO) mice (NUR77 KO). This observation provides one with an intriguing genetic tool, which can be used to delineate the developmental relationship of blood and tissue Ly6C⁺ and LyC6⁻ cells. NUR77 is a transcription factor that belongs to the nuclear receptor superfamily. In the absence of NUR77, blood LyC6⁻ monocytes are generated in a greatly diminished number. Using NUR77 KO animals as reverse lineage tracers in which one monocyte subset is missing due to gene deletion allows one to ask the following questions: how is the development of tissue Ly6C⁺ and Ly6C⁻ MFs perturbed in the absence of Ly6C⁻ monocytes? Does the development of tissue Ly6C⁺ and Ly6C⁻ MFs mirror that of blood monocyte subsets?

To find answers to these questions, we used a mouse model in which sterile inflammation is caused by a single cardiotoxin injection into skeletal muscle (17, 18). In this model, muscle injury elicits massive MF accumulation and sequential generation of Ly6C⁺ and Ly6C⁻ MFs at the injury site. Compared with other inflammation models that are characterized by monocyte infiltration, the muscle regeneration model is unique, because a complete restoration of tissue structure and function ensues monocyte infiltration and MF differentiation. This is made possible by the active contribution of Ly6C⁺ and LyC6⁻ MFs. The main question is what is the ancestry of these Ly6C⁻ tissue MFs? Are the Ly6C⁻ cells derived from a Ly6C⁺ to Ly6C⁻ differentiation or from Ly6C⁻ monocyte infiltration? By examining the dynamics of Ly6C⁺ and Ly6C⁻ MF development in this model, we found that tissue Ly6C⁻ MFs were generated in wild-type (WT) and NUR77 KO animals at the same numbers and with the same dynamics. This result implicates that the development of Ly6C⁺ and Ly6C⁻ tissue MFs is markedly different from that of blood monocyte subsets. Although NUR77 regulates blood Ly6C⁻ monocyte development, it is not required for tissue Ly6C⁻ MF differentiation. Our results also indicate that in the used experimental model tissue Ly6C⁻ MFs are not generated from an infiltrating pool of blood Ly6C⁻ monocytes but are probably derived from tissue Ly6C⁺ MFs.

Nr4A1^−/− and WT C57BL/6J mouse strains were bred under specific pathogen-free conditions and used for experiments in accordance with Hungarian (license 21/2011/DEMÁB) and European regulations. Experiments were conducted on adult (2- to 6-mo-old) male mice.

Mice were anesthetized with isoflurane, and 50 μl cardiotoxin (12 × 10⁻⁶ mol/L in PBS) was injected in the tibialis anterior (TA) muscle. Muscles were recovered for flow cytometry analysis at day 1, 2, or 4 postinjury or for muscle histology at day 8 postinjury.

Muscles were removed and snap frozen in nitrogen-chilled isopentane (−160°C). The 8-μm–thick cryosections were stained with H&E.

Fascia of the TA was removed. Muscles were dissociated in RPMI 1640 containing 0.2% collagenase B (Roche Diagnostics) at 37°C for 1 h, filtered, and cells were counted. CD45⁺ cells were isolated using magnetic sorting (Miltenyi Biotec), treated with FcγR blocking Abs, and stained with a combination of PE-conjugated anti-Ly6C Ab (eBioscience) and allophycocyanin-conjugated F4/80 Ab. In each experiment, both genotypes were processed to minimize experimental variation. Cells were analyzed with either a cell sorter (BD FACSAria III) or a flow cytometer (BD FACSCalibur; Fig. 2E). Living cells were gated according to their forward light scatter (FSC) and side light scatter (SSC) properties. This was combined with a gate on the singlet cell population according to the signal width of FSC and SSC parameters. Labeling of cell surface receptors Ly6C and F4/80 on living singlet cells was quantified.

MF subsets were sorted from day 2 postinjury muscles with a FACSAria III sorter, and total RNA was isolated with TRIzol reagent, according to the manufacturer’s recommendation. 20 μg glycogen was added as carrier for RNA precipitation. A mouse strain carrying an irrelevant genetic modification (Cre recombinase expression without floxed alleles, strain: B6.129P2-Lyzs^tm1(cre)Ifo/J) was used for MF sorting and RNA isolation.

Global expression pattern was analyzed on Affymetrix GeneChip Mouse Gene 1.0 ST arrays. Ambion WT Expression Kit (Life Technologies) and GeneChip WT Terminal Labeling and Control Kit (Affymetrix) were used for amplifying and labeling 150 ng total RNA. Triplicate samples were hybridized at 45°C for 16 h, and then standard washing protocol was performed using Affymetrix GeneChip Fluidics Station 450. The arrays were scanned on GeneChip Scanner 7G (Affymetrix). Microarray data (data access: GSE44057, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE44057) were analyzed with GeneSpring 12 GX software (Agilent BioTechnologies). Affymetrix CEL files were normalized with Robust Multichip Analysis algorithm and median normalization.

H&E stained muscle sections were recorded with a Nikon E800 microscope at original magnification ×20 connected to a QIMAGING camera. For each TA, myofibers in at least three fields randomly chosen in the entire injured area were counted, yielding a total of 3131 (Nr4a1^−/−) or 2690 (WT) counted myofibers. Quantitative analysis of muscle regeneration (necrotic, phagocyted, basophilic, and centrally nucleated myofibers) was performed using the Image J software and was expressed as a percentage of the total number of myofibers.

Blood was taken from the inferior vena cava into EDTA-containing tubes and hemolysed with RBC lysis buffer (BioLegend), according to the manufacturer’s recommendation. Following FcγR blocking, cells were stained with a combination of PE-conjugated anti-CD115 (eBioscience) and PerCP-Cy5.5–conjugated anti-Ly6C (eBioscience). A FSC/SSC gating strategy was used to include monocytes only in quantification of monocyte subset ratio.

All experiments were performed using at least three different samples. Student t tests were performed and a p value < 0.05 was considered significant. Mean and SD values are shown in graphs.

In this study, we aimed to reveal the developmental history of tissue Ly6C⁺ and LyC6⁻ MF populations and to compare the differentiation of blood and tissue Ly6C⁺ and LyC6⁻ myeloid cells. We used a mouse model of skeletal muscle injury/regeneration in which sterile inflammation is caused by a single cardiotoxin injection, which causes massive muscle fiber death. The sterile inflammation is accompanied by the generation of MF populations in the regenerating muscle comprising first Ly6C⁺ F4/80^low and then LyC6⁻ F4/80^high subsets. This model has two features that make it especially useful for the study of Ly6C⁺ and LyC6⁻ myeloid cell differentiation. On the one hand, the model is characterized by a rapid and robust MF infiltration, accumulation, and differentiation at the injury site, which enabled us to easily quantify cellular subsets. Additionally, the MF populations exhibit a dynamic transition of cellular phenotypes as exemplified by their cell surface markers. The major features of this experimental model and the involved cell types are presented in Fig. 1A. We isolated hematopoietic cells from the regenerating muscle using positive CD45⁺ bead selection. Cell separation was followed by FACS analysis with Ly6C and F4/80 double staining. The earliest stage of muscle injury (Fig. 1A) is characterized by a robust infiltration of Ly6C⁺ F4/80^low MFs and LyC6^int F4/80⁻ neutrophils to the site of the injury (18). The initial Ly6C⁺ F4/80^low MF population rapidly disappears to the benefit of an increasing population of LyC6⁻ F4/80^high population during the course of regeneration, whereas the neutrophil infiltration is, at least partially, cleared from the muscle. Four days after muscle injury, at a stage that is characterized by active muscle regeneration, the Ly6C⁺ population is almost completely supplanted by the Ly6C⁻ cells.

We used this muscle injury/regeneration model to answer the question as to whether the defect in the development of blood monocyte subsets is mirrored in the development of tissue MF development. When comparing the characteristic cell types of regenerating muscle of WT and NUR77 KO animals (Fig. 1A), it was strikingly evident that NUR77 KO animals appeared to have an unaltered LyC6⁻ development. This finding was surprising in light of the fact that we detected robust expression of NR4A subfamily members (NR4A1, NR4A2, and NR4A3, or Nur77, Nurr1, and Nor1, respectively) in the Ly6C⁺ and LyC6⁻ MF subsets (Fig. 1B). In fact, we found that NR4A1 and NR4A2 were among the top 25 transcription factors when all transcription factors were ranked according to their mRNA expression level in the Ly6C⁺ MF subset. The finding that NUR77 KO animals had apparently normal LyC6⁻ MF development even though these animals were deficient in generating a normal LyC6⁻ blood monocyte compartment meant that the development of blood and tissue Ly6C⁺ and LyC6⁻ myeloid cells is likely to be different. The sole fact that NUR77 KO animals generated LyC6⁻ MFs did not exclude the possibility, however, that the dynamics of the generation of LyC6⁻ MFs was not the same in WT and NUR77 KO animals. Therefore, we set out to characterize cellular transitions and subsets in regenerating muscle in detail. We chose days 1, 2, and 4 of muscle regeneration for detailed longitudinal analysis. Day 2 muscle samples in WT animals contained comparable ratio of Ly6C⁺ and LyC6⁻ MFs, and, therefore, day 2 samples could report either a positive or negative change in the transition of MF subtypes. In contrast, day 4 samples in WT animals contained predominantly LyC6⁻ cells; therefore, this time point was deemed to be the optimal one to report on the overall efficiency of LyC6⁻ cell generation. Fig. 2A shows that, at day 2 of muscle regeneration, the number of hematopoietic cells (including neutrophils, Ly6C⁺ and LyC6⁻ MFs) isolated from the regenerating muscle was the same in WT and NUR77 KO animals. The fact that the overall number of hematopoietic cells was identical made it possible to directly compare the number of Ly6C⁺ and LyC6⁻ MFs in the regenerating muscle (Fig. 2B, 2C) between genotypes. We found that, although a slight tendency for larger variation in the numbers of the MF subsets in NUR77 KO samples was seen, Ly6C⁺ F4/80^low and LyC6⁻ F4/80^high MFs were generated at the same ratio in WT and NUR77 KO animals (22.7 ± 3.9% versus 24.3 ± 4.5% LyC6⁻ F4/80^low cells and 16.1 ± 2.4% versus 16.4 ± 5% LyC6⁻ F4/80^high cells in WT and NUR77 KO animals, respectively). This finding formally suggested that the process of LyC6⁻ MF differentiation was unaffected in NUR77 KO animals. This is in stark contrast with the fact that NUR77 KO animals had a severely diminished pool of blood LyC6⁻ monocytes (59.63 ± 1.6% versus 6.4 ± 4.2% LyC6⁻ CD115⁺ cells in WT and NUR77 KO animals, respectively) (Fig. 2D), as it was reported earlier (16). The number of hematopoietic cells in the regenerating muscle is the sum of cell infiltration, cell division, cell egress, and cell death. It cannot be excluded, therefore, that the identical size of the infiltrating cell pool at day 2 of regeneration was not a sign of an unaltered cell infiltration/development, but in fact a result of a complex perturbation in several of the above processes. Such a complicated alteration could, by pure chance, give rise to a cell pool of seemingly same size but markedly different composition. To ascertain that cellular dynamics were not different in NUR77 KO animals compared with WT mice, we analyzed cellular composition of infiltrating cells in regenerating muscle (Fig. 2E) at day 1. We found no difference in the composition of cellular infiltration (comprising neutrophils and Ly6C⁺ MFs) at the early stages of inflammation/regeneration (70.1 ± 6.6% versus 67 ± 8.5% neutrophils and 21 ± 4.16% versus 25.6 ± 7.9% Ly6C⁺ MFs in WT and NUR77 KO animals, respectively). Based on the above findings, we concluded that the same type of MFs was generated at the same number from a virtually identical cell infiltrate in WT and NUR77 KO animals during the early stages of regeneration. Finally, we wanted to see whether the overall efficiency of the generation of LyC6⁻ cells was identical in WT and Nur77-deficient animals. At day 4 of muscle regeneration, the MF infiltration in WT animals was indistinguishable from the infiltration detected in NUR77 KO mice (Fig. 2F). In short, the infiltration, development, and differentiation of Ly6C⁺ and LyC6⁻ tissue MFs were found to be identical in the two examined mouse strains that showed characteristically different blood LyC6⁻ monocyte compartments.

It has been reported that the LyC6⁻ blood monocyte compartment in NUR77 KO animals is not only generated in much smaller number, but the existing LyC6⁻ monocytes were shown to have altered morphology and propensity for cell death, which suggests that these cells had functional defects. In this study, we wanted to ask whether either the Ly6C⁺ or the LyC6⁻ MF compartment, which appeared to be generated in normal numbers in NUR77 KO animals, did show some functional defect. MFs are critical regulators of muscle regeneration, although the underlying molecular mechanisms responsible for the proregenerative role of tissue MFs are poorly understood. These functions are likely to involve differential immunomodulatory, proangiogenic, and/or myogenic roles (18, 19). In experiments in which peripheral monocytes are ablated, a total impairment in muscle regeneration is observed. Limited monocyte infiltration into injured muscle triggers a delay in regeneration, which can be detected by histological scoring of H&E-stained muscle sections. Early in regeneration, damaged myofibers undergo cell death and are cleared up by the phagocytic action of MFs. These early stages of regeneration, therefore, can be characterized by a high number of necrotic and phagocyted fibers. Later, new regenerating myofibers with a basophilic staining pattern arise and then grow into bigger myofibers, which are characterized by central nuclei. In this study, we used histological examination of regenerating muscle as a functional measure of Ly6C⁺ or LyC6⁻ MF functions during muscle regeneration. When regenerating muscle samples from WT and NUR77 KO animals were subjected to histological analysis (Fig. 3A) at day 8 after muscle injury, no visual difference in the overall regeneration efficiency and the generated cell types was detected. Same pattern of regeneration was observed in both cases, with the presence of numerous regenerating myofibers and some remaining interstitial inflammatory cells. As a further proof for the lack of an effect of Nur77 deficiency in muscle-derived MFs, we counted the numbers of different types of fibers present during regeneration. We used the ratio of the necrotic/phagocyted fibers versus basophilic/regenerating fibers as a simple readout of the efficiency and overall rate of regeneration. As no difference was observed in the involved cell types in WT and NUR77 KO animals (Fig. 3B), we concluded that Nur77 deletion in muscle-derived MFs did not alter normal muscle regeneration, and, therefore, Nur77-deficient MFs, especially the LyC6⁻ subset of them, are not only generated in the usual number but show no functional alterations in this experimental model (Fig. 4).

In this study, using a model of spontaneous resolution of sterile inflammation, we were able to provide answers to several unresolved questions of myeloid cell development. The first definitive finding is that the development of Ly6C⁺ and LyC6⁻ tissue MFs, at least within the context of the used experimental system, is clearly different from that of blood Ly6C⁺ and LyC6⁻ monocytes. Recent evidence (15) supports a model in which Ly6C⁺ blood monocytes serve as obligatory precursors for the generation of LyC6⁻ monocytes, although other results argue for the scenario in which blood LyC6⁻ monocytes are derived from bone marrow progenitors without the direct involvement of Ly6C⁺ blood monocytes (16). Whichever model is correct, it is certain that Ly6C⁺ monocytes are generated normally, whereas LyC6⁻ blood monocytes are diminished in Nr4A1^−/− animals. The experimental model used by us is characterized by the generation of large numbers of Ly6C⁺ tissue MFs. In this study, however, the parallel to monocyte development ends, because in a genetic context in which the generation of LyC6⁻ monocytes is blocked, the generation of LyC6⁻ tissue MFs is unperturbed. This points to the existence of a distinct or an alternative mechanism by which LyC6⁻ tissue MFs are generated. The second question our study gave insight about is the relationship between blood and tissue Ly6C⁺ and LyC6⁻ cells. What is the origin of tissue Ly6C⁺ and LyC6⁻ cells in the current experimental model? Several recent papers gave overlapping, yet distinct answers for the question of contribution of Ly6C⁺ or LyC6⁻ blood monocytes to tissue MFs or DCs. Blood Ly6C⁺ monocytes can extravasate and infiltrate inflamed tissue in a number of different inflammation models (1). It has been suggested that these infiltrating Ly6C⁺ monocytes could give rise to inflammatory Ly6C⁺ and then to LyC6⁻ MFs (5). Other models described that Ly6C⁺ monocytes can differentiate into DCs (20, 21). Another study, utilizing a colon inflammation model, found that infiltrating Ly6C⁺ monocytes gave rise to Ly6C⁺ inflammatory MFs and a LyC6⁻ population with functional features of DCs (22). In contrast, at least in certain models of tissue damage or infection, LyC6⁻ monocytes were also able to infiltrate the injury site, and therefore these cells were most likely the precursors to LyC6⁻ tissue MFs in these inflammatory contexts (9, 23). In short, monocyte/MF differentiation appears to be heavily influenced by tissue environment and the choice of inflammatory models. Which of the above models parallels our findings? In our model, LyC6⁻ tissue MFs are generated in the almost complete absence of LyC6⁻ monocytes. This observation has two consequences, as follows: in this model system the contribution of LyC6⁻ monocytes to infiltration is probably minimal or negligible and therefore, 1) tissue LyC6⁻ MFs are not likely to be generated from LyC6⁻ monocytes, but 2) most probably from Ly6C⁺ tissue MFs (Fig. 4). It is important to stress that in our experimental model Ly6C⁻ cells accumulate rapidly and in high numbers. These two features make the model in which LyC6⁻ MFs are generated from infiltrating LyC6⁻ monocytes highly improbable, regarding the unaltered rate of LyC6⁻ MF development when the LyC6⁻ monocyte subset is so critically impacted. Our results are in line with earlier studies (based on bead-tracing experiments and monocyte depletion/MF recovery kinetics) (18), suggesting that tissue Ly6C⁺ MFs in injured muscle could be direct precursors of tissue LyC6⁻ MFs. It is also important to note that the Ly6C⁻/F4/80^high cells that accumulate during muscle repair appear to have tissue regenerative functions, as their depletion at late phase of muscle regeneration impaired the growth of regenerating myofibers (18). How general are our findings relative to other subtypes of monocyte-derived myeloid cells? The skeletal muscle regeneration model used by us is unique in the sense that a complete resolution of a sterile inflammation is linked to a complete restoration of tissue function. The ancestry and development of Ly6C⁺ and LyC6⁻ MFs in other experimental systems can be remarkably different. Regardless, we propose that the strategy used in this study, that is, using NUR77 KO animals as a reporter strain in which the LyC6⁻ monocyte compartment is diminished, appears to be an excellent method in the toolbox of myeloid cell biology that could complement currently used experimental strategies that use complex manipulations of the immune system, such as adaptive transfer or lineage tracing by CX3CR1^GFP/+ cells. Using this strategy, the development of noncanonical myeloid cell types, such as myeloid suppressor cells or the angiogenic Tie2⁺ cells, could be revisited.

Finally, our results also revealed that NUR77, which is an important regulator of blood LyC6⁻ monocyte development, is not a critical regulator of tissue MF development and functions in the muscle injury/regeneration experimental system. It should be noted that NR4A3 (NOR-1), a closely related member of the nuclear receptor family, has been shown to be coregulated and to have overlapping target genes with NUR77 in myeloid precursor cells. In fact, null mutation of both Nr4A1 and Nor1 was necessary to induce a rapidly lethal myeloid leukemia in mice. It cannot be excluded, therefore, that the apparent lack of consequence of Nur77 deletion in MF development is due to compensatory mechanisms provided by other NR4A members, for example, by NOR1. It could be reasoned that, similarly to monocyte development, there must exist transcriptional regulators of MF development that regulate cell fate in a cell subset–specific way. This necessitates further inquiry into the transcriptional regulators of tissue MF development.

The authors have no financial conflicts of interest.