Macrophages modulate fibrosis during newt lens regeneration

Previous studies indicated that macrophages play a role during lens regeneration in newts, but their function has not been tested experimentally. Here we generated a transgenic newt reporter line in which macrophages can be visualized in vivo. Using this new tool, we analyzed the location of macrophages during lens regeneration. We uncovered early gene expression changes using bulk RNAseq in two newt species, Notophthalmus viridescens and Pleurodeles waltl. Next, we used clodronate liposomes to deplete macrophages, which inhibited lens regeneration in both newt species. Macrophage depletion induced the formation of scar-like tissue, an increased and sustained inflammatory response, an early decrease in iris pigment epithelial cell (iPEC) proliferation and a late increase in apoptosis. Some of these phenotypes persisted for at least 100 days and could be rescued by exogenous FGF2. Re-injury alleviated the effects of macrophage depletion and re-started the regeneration process. Together, our findings highlight the importance of macrophages in facilitating a pro-regenerative environment in the newt eye, helping to resolve fibrosis, modulating the overall inflammatory landscape and maintaining the proper balance of early proliferation and late apoptosis.


Background
A popular hypothesis, fueled by the observations that regeneration is gradually lost through ontogeny and evolution, suggests that adult mammals have developed a more robust adaptive immune response at the expense of regenerative capabilities (1).However, NOD/SCID mice that exhibited T cell de ciency but maintained macrophage numbers failed to regenerate their hearts at neonatal stages and demonstrated signs of severe brosis, suggesting that modulating adaptive immunity alone is not su cient for successful regeneration (2).Furthermore, the theory of inverse relationship between regeneration and immune pro ciency fails to explain why several mammals are capable of epimorphic-like regeneration during adulthood (3,4,5).An alternative theory suggests that the interaction between immune cells and the local microenvironment in uences the regeneration outcome, rather than the absence or presence of special immune and regenerative cells (6).
Macrophage involvement during tissue regeneration has been a subject of intense discussion in recent years (7,8,9,10).There is now increasing evidence that macrophages play a necessary role during limb, n, tail, spinal cord, and heart regeneration in a variety of species (11,12,13,14,15,16,17,18,19,20).The regenerative process in the aforementioned tissues and organs is complex, requires the integration of multiple cell populations, or is not easily accessible for visualization (21,22,23,24).This added level of complexity makes it di cult to extrapolate immune mechanisms in a regenerative context.On the other hand, the case of lens regeneration in newts involves the transdifferentiation of a single cell type, the iris pigment epithelial cell (iPEC) into lens cells (25,26,27,28,29,30).The elegant simplicity of this process serves as a great platform to uncover the mechanism by which macrophages promote or interfere with scar-free healing and regeneration.
In the past, studies utilizing electron microscopy technology recognized that macrophages, which were rst called "special amoeboid cells" due to their morphology, migrated inside the iris epithelium at 3 days post-lentectomy (dpl) and phagocytosed melanosomes that were discharged from iPECs during the dedifferentiation process (31,32,33,34).In the present study, we developed and characterized a new Pleurodeles waltl transgenic line with eGFP-labeled macrophages to enable the study of macrophages during the regeneration process.Furthermore, we characterized the immune landscape of the iris after injury and investigated the role of macrophages in directing the early and late stages of lens regeneration.
We demonstrated that macrophage depletion modulates the wound healing response and affects the regenerative outcome.In addition, we demonstrated that macrophages returning into the deleterious wound lesion were unable to resolve the in ammatory and brotic environment.However, a secondary injury alone or the addition of broblast growth factor 2 (FGF2) restarted scar resolution and the regeneration processes.

Animal husbandry and ethical statement
Iberian Spanish newts, Pleurodeles waltl, and red spotted eastern newts, Notophthalmus viridescens were used in this study.Spanish newts were bred and grown in our colony at Miami University, whereas red spotted eastern newts were wild caught.Handling and surgical procedures were performed following guidelines by the Institutional Animal Care and Use Committee at Miami University.Adult Pleurodeles waltl wildtype newts born and raised in captivity for generations were used for transgenesis in the aquatic facilities of Karolinska Institutet, Stockholm.Handling, breeding, transgenesis and the other experimental procedures performed in Stockholm were done according to both Swedish and European regulations.All animals were raised according to previously established husbandry guidelines (35).
It's important to note, that the lens growth rates and speed of regeneration can vary between different newt species and between different ages of the same species (28,36,37).For this reason, we refer to lens regeneration stages (as de ned in (36)) in addition to days post-lentectomy throughout the manuscript.
Generation of Tol2-mpeg1:eGFP-polyA plasmid A vector containing a 1.86kb fragment of the zebra sh mpeg1 promoter driving eGFP as previously described (38), was a kind gift from Enrique Amaya.The mpeg1 fragment was ampli ed by PCR (mpeg1 F 5'-TTGGAGCACATCTGAC − 3'; mpeg1 R 5'-TTTTGCTGTCTCCTGCAC-3') and subcloned into pBSII-SK-mTol2 upstream of the coding region of EGFP.SV40 polyA was used as termination signal.The resulting plasmid was puri ed using cesium chloride preparation to avoid potential contaminants that may negatively impact on transgenesis e ciency or newt survival.The plasmid was subsequently puri ed using Qiagen Maxiprep according to manufacturer's instructions and then resuspended in ultrapure distilled water.

Lentectomy, iridectomy and EdU injections
Animals were anesthetized by whole body submersion in 0.1% ethyl 3-aminobenzoate-methane Sulfonic acid solution (MilliporeSigma) diluted in amphibian phosphate-buffered saline (APBS :1x PBS plus 25% dH2O).Once the animal was anesthetized, a scalpel was used to make a slit in the cornea and the entire lens was carefully removed with ne tweezers (46).Clodronate or PBS liposomes (Encapsula Nano Science, # CLD-8901) were injected intraocularly into the vitreous cavity of the newt eye using a prepulled glass needle (20µm tip diameter) attached to a microinjector (MicroJect 1000A, BTX, Harvard Apparatus) set at 10.5psi.For iridectomy experiments, the apical region of the dorsal iris (and regenerating lens from PBS-liposomes treated eyes) was surgically removed at 60 dpl by re-opening the cornea with a scalpel and removing a piece of the iris using scissors and tweezers.After all surgical procedures, the animals were allowed to recover from anesthesia and return to appropriate housing containers where they were monitored carefully for the duration of the experiments.For cell proliferation studies, EdU (Invitrogen, #C10338) was injected intraperitoneally 24 hours prior to collection at 10 µg/g of body weight.All animals used for these experiments were post-metamorphotic, juveniles at 6-8 months old.

FGF2 experiments
Heparin-coated polyacrylamide beads (Sigma, #H-5263) were washed in APBS and incubated either with 0.25µg/µl bFGF (R&D Systems, #133-FB) or APBS (vehicle control) overnight at 4°C.Heparin beads were carefully inserted with tweezers into the vitreous chamber of the eye, between the dorsal and ventral iris following lens removal.All animals used for these experiments were post-metamorphotic, juveniles at 6-8 months old.
Notophthalmus viridescens RNAseq, transcriptome assembly, and differential expression analysis Dorsal iris tissues were collected from adult Notophthalmus viridescens from intact animals (no lentectomy), as well as animals 6 hpl, 1 dpl, and 4 dpl.Three biological replicates were used per time point, with each biological replicate containing bilateral dorsal irises from 3 or 4 animals pooled together.The iris tissues were placed in 500µL of cold TRI Reagent (Zymo, R2050-1-50), vortexed, and stored at -80°C until RNA extraction.RNA isolation was performed via extraction with 0.2 volumes chloroform followed by processing of the aqueous phase with the Direct-zol RNA Microprep Kit (Zymo, R2060) according to manufacturer's instructions, including an in-column DNase I treatment.RNA integrity was assessed with the Agilent RNA 6000 Pico Kit (Agilent, 5067 − 1513) and quanti cation performed with the Qubit RNA HS Assay Kit (ThermoFisher, Q32852).Reverse transcription and library preparation were carried out with the Zymo-Seq RiboFree total RNA library Kit (Zymo, R3000) using 108 ng input RNA per reaction.RiboFree depletion was applied for 4 hours to deplete overrepresented transcripts, and nal libraries were ampli ed with dual indexes (Zymo, D3096) for 13 PCR cycles.Sequencing was performed at the Novogene sequencing core on the Illumina NovaSeq 6000 to approx.100 million paired-end reads per sample.
The rst 10 base pairs of sequence read2 were hard trimmed to remove low-complexity bridge sequences introduced during library preparation.The reads were then quality and adapter trimmed with Trim Galore using parameters -q 5 --length 36 --stringency 1 -e 0.1 (47,48).rRNA reads were depleted by aligning trimmed reads against the Silva rRNA database using Bowtie2 (49,50).Cleaned reads were prepared for transcriptome assembly using the Trinity assembler pre-processing script insilico_read_normalization.pl with parameters --seqType fq --JM 1450G --max_cov 30 --pairs_together --SS_lib_type RF -CPU 24 --PARALLEL_STATS --KMER_SIZE 25 --max_CV 10000 --min_cov 2 (51).Trinity assembly was performed on each condition individually with Trinity using parameters --seqType fq --max_memory 1450G --SS_lib_type RF --CPU 24 --min_contig_length 200 --monitoring --min_kmer_cov 2 --no_normalize_reads. Individual assemblies were merged into a nal transcriptome with the DRAP pipeline runMeta tool and parameters -strand RF --mapper bwa --length 200 --type contig --coverage 0,1,10 --write (52).Assembled transcripts were annotated against the NCBI non-redundant database using an implementation of blastx and the functional annotation module in the OmicsBox software environment (53).Final transcript abundance was estimated with the Salmon alignment tool using the parameters -l ISR --numGibbsSamples 20 --seqBias --gcBias --reduceGCMemory -d (54).Differential transcript abundance testing was performed in the R environment using the SWISH Fishpond work ow (55).Expression values displayed in manuscript are the log-transformed Transcript Per Million (TPM) values, averaged across inferential replicates.In the case of transcripts with multiple detected isoforms, the transcripts displayed in the heatmap represent the transcript assigned the lowest E-value determined by blastx.KEGG pathway enrichment analysis was performed using the Combined Pathway Module inside the OmicsBox environment, with signi cance values calculated using maSigPro time-course analysis (56, 57).For combined pathway analysis, parameters were set as Keep most speci c pathways = TRUE and Blast expectation value = 0.001, and KEGG orthologs were linked through the EggNog mapper (58).A two tailed Fisher's test was applied to pathways and FDR cut off was applied at 0.05.

Pleurodeles waltl RNAseq and differential expression analysis
Dorsal iris tissue was dissected from 13-month-old newts using intact or lentectomized animals.Tissue from 4 animals were pooled for each biological replicate, collected in triplicate.Samples were collected into cold TRI Reagent (Zymo, R2050-1-50), vortexed, and stored at -80°C until RNA extraction, as described above.RNAseq libraries were prepared using 72-130 ng of isolated RNA, using NEBNext® Ultra™ II Directional RNA Library Prep with Sample Puri cation Beads (NEB, E7765S) and NEBNext® Poly(A) mRNA Magnetic Isolation Module (E7490S).Indexing was performed using NEBNext® Multiplex Oligos for Illumina® (NEB, E7500/7710/7730).Samples were pooled and sequenced on a lane of NovaSeq 6000 at the Novogene Sequencing Core using paired-end, 150 base pair reads to a minimum depth of 29 million read pairs per sample.
Reads were quality trimmed using Trim Galore with parameters --stringency 3 --paired --length 36 (47,48).Cleaned reads were aligned to the Pleurodeles waltl genome using the STAR aligner with two-pass alignment to insert splice junctions (59,60).Aligned reads were assembled into transcripts using Stringtie and spliced transcripts were extracted using gffread (61, 62).Assembled transcripts were indexed using the Salmon quanti cation tool, using the parameter -k 31 and providing genomic sequence as decoy (54).
Transcript expression was quanti ed using salmon with parameters -l ISR --validateMappings -p 12 --numGibbsSamples 20 --seqBias --gcBias -d.Samples with alignment rates > 80% were included in downstream analysis, resulting in the exclusion of one sample (intact replicate 3).Expression values displayed in manuscript are the log-transformed Transcript Per Million (TPM) values, averaged across inferential replicates.Assembled transcripts were annotated against the NCBI non-redundant database using an implementation of blastx and the functional annotation module in the OmicsBox software environment (53).

RT-qPCR design and analysis
Whole eyes were enucleated at the indicated time points and placed in 500µL of TRIzol and stored at -20°C.The tissues were then homogenized mechanically using pellet pestles and centrifuged to remove debris.S1.TAII70, was used as a housekeeping gene using primers published previously (64).The comparative Ct method was used to determine relative gene expression levels compared to the housekeeping gene.The primers and qPCR reactions were validated following qPCR MIQE guidelines (65).Four to eight biological samples were used per condition.
In order to improve data conformance to statistical modeling assumptions, a natural log transformation was applied to relative mRNA values.For time course analysis of intact eyes through 30 dpl eyes (Figs.4A), a two-way block ANOVA was conducted, with treatment and time as the factors along with their interaction.The blocking factor was included to account for batch effects, as animals were collected in two groups.The analysis was done using the aov function in the R stats package (66).Treatment differences were estimated for each time point using the emmeans function in R, and since ve comparisons were made, the p-values were corrected by controlling the false discovery rate (67, 68, 69).
For experiments in which transcript expression was assayed at two time points (Fig. 5D), a two-way ANOVA with interaction (factors Treatment and Time) was performed using the lm function in R, part of the stats package.Analysis of treatment means and multiple comparisons correction was done as described above.For the ANOVAs, residuals were examined to check for severe assumption violations, including constant error variance and normality.For experiments with a single time point (Fig. 6E), a Welch's two-sample t-test was applied using the R function t.test, part of the stats package.All animals used for these experiments were post metamorphotic, juveniles at 6-8 months old.For more information on statistical analysis see Appendix 1.

Optical Coherence Tomography
The anterior chamber of each eye was non-invasively monitored via spectral domain optical coherence tomography (SD-OCT) as previously described (70).The animals were anesthetized prior to imaging, and 100 µL of water was applied to the cornea surface with a pipette to reduce re ection artifacts caused by dehydration during live imaging.The broadband light source was ltered to produce a low-coherence spectrum center around 850nm with frequency range near 200 MHz, which can produce an axial resolution of 2 µm and lateral resolution of 8 µm.The scanning area of the eye was 1.2 by 1.2 mm, where 500 B-Scans were collected across this range.Each B-Scan consisted of 2000 A-Scan, where each A-Scan had 2048 pixels.The nal C-Scans were rescaled and reconstructed as 500 x 500x 500 voxels.All animals used for these experiments were post metamorphotic, juveniles at 6-8 months old.

Tissue embedding and sectioning
For cryo-embedding, collected tissues were washed 2 to 3 times in PBS and xed overnight in 4% PFA (1 g paraformaldehyde in 25 mL PBS) at 4°C.For cryoprotection, animals were transferred to new tubes with 30% sucrose (in 0.1M PBS) overnight at 4°C.Tissues were placed in embedding molds, correctly positioned and tissue tek (yellow Shandon Cryochrome, Thermo Fisher Scienti c, Waltham, USA) was gently added.Embedded molds were placed at -80°C for 15 minutes and transferred to -20°C and stored until sectioned.For sectioning, a cryostat was used with an object temperature of -21°C and knife temperature of -19°C.For experiments involving whole animals, 12 µm sections were made.Sections were collected on Superfrost Plus slides and stored at -20°C.Tissue processing for the phagocytic experiments depicted in Fig. 1(E, F) were done according to Joven et al., 2018 (45): in brief, gelatinembedding and colder temperatures (-30°C) were used for sectioning.
For para n embedding, whole eyes were washed 3 times in PBS and xed overnight in 10% formalin at 4°C.For intact lenses or later time points (100 dpl) when the lens is large, eyes were xed in methanol: acetic acid (3:1 ratio) overnight at 4°C to preserve lens morphology.Para n embedded tissues were sectioned at 10µm thickness using a microtome.

Immuno uorescent staining
Slides with cryo-sectioned eyes were air-dried for 30 minutes at RT and thereafter washed 3 times with PBST (0.2% Triton in 1 L PBS) for 10 minutes.To make the tissue more accessible to antibodies, a retrieval procedure was used.Plastic containers were lled with a citrate buffer (120 µL antigen unmasking solution (Vector, Peterborough, UK) in 11.88 mL PBS and preheated in a water bath to 86°C.Slides were introduced once the solution reached 86°C and incubated for 10 minutes.Thereafter, slides were placed in a glass container with PBS at room temperature to cool down.Blocking buffer (10% goat or donkey serum in 0.2% PBST) was added to the slides for 1 hour at RT. Subsequently, primary antibodies were added to the slides and incubated overnight at 4°C.List of primary antibodies and concentrations used can be found below.Slides were washed 3 times with PBST for 10 minutes.Appropriate secondary antibodies (AlexaFluor 488 or 594 conjugates, ThermoFisher) were added to the slides and incubated for 2-4 hours at RT protected from light.Slides were washed 3 times with PBST for 10 minutes.Hoechst (1:10000 in PBS) was added for 15 minutes at RT and protected from light.Slides were washed 3 times with PBST for 10 minutes and mounted with Fluorescent Mounting Medium (Sigma, #F-4680).Immuno uorescent staining procedures for the phagocytic experiments depicted in Fig. 1(E, F) were done according to Joven et al., 2018 (45).The same protocol was followed for para n embedded tissues, with additional depara nization steps.The depara nation steps involved two xylene washes for 5 minutes, and gradually rehydration of the tissue by 1-minute washes with 100%,95%,80%70%,50%,30% ethanol followed by three PBS washes.All animals used for these experiments were post-metamorphotic, juveniles at 6-8 months old with an exception of the MPEG transgenic animals that were 3 years old.

Histology and Cytochemistry
Following depara nization, hematoxylin and eosin, and picrosirius red staining (Polysciences, #24901) were performed following manufacturer protocols.Prior to EdU and TUNEL assays, sections were depara nized, and incubated with 0.01M Sodium Citrate (pH = 6) for 15 min at 95°C for antigen retrieval.A permeabilization step was followed in which the tissue sections were washed with 1% saponin and APBST wash buffer (APBS supplemented with 0.1% Triton-X100).Slides were then incubated in EdU (Invitrogen, #C10337) or TUNEL (Roche #11684795910 or #12156792910) reaction cocktails according to manufacturer guidelines.Slides were then washed 3 times in PBST for 10 min and nuclear counterstain was achieved by incubating slides with Hoechst 33342 (Invitrogen, #H3570) or DAPI (Life technologies, #D1306) at 1:1000 dilution.Slides were then washed in PBST 3 times and mounted with uorescent mounting media (Sigma, #F-4680).

Microscopy and imaging analysis
Live imaging of larvae, screenings and time-lapse movies were performed using either a Zeiss Axiovert 200 M inverted microscope or a LeicaM205 FCA stereo uorescence microscope equipped with a Leica DMC 6200 camera.Figure 1A, B images were produced by processing a Z-stack acquisition using the function "Extended Depth of Focus" of LAS X version 3.7.4software.Confocal images were obtained using Zeiss 700 and 710 Laser Scanning Confocal System Carl Zeiss, Gottingen, Germany).Z-stack con gurations (Fig. 1E: 26 images at 1µm intervals and 2048 x2048 size Fig. 1F: 10 images at 1µm intervals and 1024x1024 size) were used to obtain high resolution images using ZEN 2012 Browser (Carl Zeiss, Gottingen, Germany).Fluorescent imaging was performed using a Zeiss Fluorescence Stereomicroscope Axio Zoom.V16 (Carl Zeiss, Oberkochen, Germany).ImageJ was used for image analysis.

EdU + cell quanti cation and statistical analysis
To determine whether the number of EdU + cells in proportion to the total number of HOECHST + cells give evidence of a difference between groups, we used Negative Binomial regression on EdU + cell count with the two-level treatment as the predictor (PBS Liposome vs. Clodronate Liposome, Fig. 3E; Clodronate/PBS bead vs. Clodronate/FGF bead, Fig. 5B) and the number of HOECHST + cells as the offset (71,72,73).
This model estimates the ratio of the mean EdU count to the total number of HOECHST + cells for treatment versus control.Note that we used Negative Binomial regression because the measurement of interest is the number of EdU cells in proportion to the total number of HOECHST + cells.These are counts rather than continuous numeric measurements, so an analysis based upon a count distribution is more appropriate than the more commonly used normal distribution-based procedures like the t-test.We used Negative Binomial regression rather than Poisson regression because there was evidence of overdispersion in the datasets.Because there were 4 eyes assigned to each treatment, we used a total of 8 EdU counts (and the associated HOECHST counts) to t the model.Note that the Negative Binomial procedure is only approximate, since we have small sample sizes, and that by using the number of HOECHST + cells as an offset, we are conditioning on their value.All animals used for these experiments were post metamorphotic, juveniles at 6-8 months old.For more information on statistical analysis see Appendix 1.

Results
mpeg1:eGFP MHY−SIMON : A new transgenic newt reporter line for macrophages Macrophage expressed gene 1 (mpeg1) has been successfully employed for driving uorescent reporters to label macrophages in several species (38,74).Seeking to employ a similar approach for labeling macrophages in Pleurodeles waltl, we created a transgenic line in which the orthologous mpeg1 promoter drives the expression of an eGFP-encoding sequence.We selected the positive offspring of the founder animals to obtain our rst generation (F 1 ) of the Pleurodeles waltl transgenic line tgTol2(Dre.mpeg1:eGFP)MHY/SIMON (referred as mpeg1:GFP from now on).To test the stability of our transgenic line, we screened all the mpeg1:GFP embryos from generation F 1 onwards for uorescent signals.In the positive offspring, eGFP + cells were found spread through the body of the animals, likely representing both resident and circulating macrophages as well as microglia cells in the brain (Fig. 1A, Video 1).Like macrophages and microglia in other species, morphologically the eGFP + cells showed a variety of shapes, including spherical, amoeboid and dendritic (Fig. 1B).Notably, macrophages in many species are notoriously auto uorescent (75,76,77).To ensure that the endogenous eGFP signal was not a result of auto uorescence, we performed immunostainings against GFP.We observed 100% colocalization between endogenous GFP and antibody-derived GFP signal, indicating that the endogenous uorescence indeed corresponds to the expression of the transgene (Supplementary Figure S1 A-C, P).To determine whether macrophages are speci cally labelled in different tissues in our mpeg1:GFP transgenic line, we tested the colocalization of eGFP + cells with an established macrophage marker, colony-stimulating factor 1 receptor (CSF1R), in eye tissues.We observed eGFP + cells and CSF1R colocalization in the cornea, vitreous chamber, and iris stroma of newt eyes (Fig. 1C).We then performed immunostainings against two other well-established markers of macrophage populations, F4/80 and Lplastin.F4/80 is a glycoprotein expressed on the cell surface of several subtypes of mature macrophages such as microglia, Langerhans cells, and resident populations in the heart, kidney, and connective tissue (78).This marker is not expressed on the cell surface of all macrophages and the levels of antigen expression differ depending on the environment in which the macrophage is found in mice (78).In agreement with these studies (78), we observed that 41.4% of eGFP + cells from several tissues in the newt body (tail, trunk, head) were F4/80+, indicating that this model allows identi cation of mature macrophages (Supplementary Figure S1 D-I, P).L-plastin is an actin-bundling protein also expressed by macrophages (79).Notably, we also observed that a signi cant fraction of the eGFP + cells (21.6%) coexpress L-plastin, further indicating that this transgenic model enables labeling of macrophage populations (Supplementary Fig. 1J-O, P).Furthermore, we tested in vivo the phagocytic nature of the eGFP + cells by examining the ability of eGFP + cells to phagocytize glucan-encapsulated siRNA particles (GeRPs).GeRPs have been shown to be selectively incorporated by macrophages in several animal models and tissues (41,42,43).We rst injected GeRPs containing rhodamine in the cerebrospinal uid through intraventricular injection.Right after the injection, we noticed the rst particles being approached and encapsulated by eGFP + cells in the central nervous system (CNS) (Fig. 1D, Videos 2, 3).The tissue analysis 20 hours post-injection showed that GeRPs had been phagocyted by microglia and border associated macrophages along the CNS (Fig. 1E).When we injected GeRPs intraperitoneally (Fig. 1D'), we found that, consistent with the previous experiment, GeRPs had been phagocytized locally by eGFP + cells situated in the intraperitoneal cavity (Fig. 1F).
Altogether, these results show that the new mpeg1:GFP Pleurodeles waltl transgenic line labels phagocytic macrophages in vivo.

Macrophages accumulate transiently in the newt eye after lentectomy
To characterize the spatiotemporal recruitment of macrophages after lens removal, we analyzed eyes from mpeg:eGFP transgenic animals.Samples were collected prior to lentectomy, at 6 hours postlentectomy (hpl), and 4-, 10-, 15-, and 30-dpl.Few eGFP + cells were observed in the intact eyes, with most of them located at the corneal epithelium (Fig. 2A i).At 6 hpl, eGFP + cells were found in the cornea epithelium and inside the blood vessels of the iris (Fig. 2A ii).At this time, the slit that was made in the cornea during the surgical removal of the lens was not yet closed.By 4 dpl, most macrophages were found around the wound area of the cornea and in the anterior eye chamber near the dorsal and ventral irises (Fig. 2A iii).By 10 dpl, the corneal incisions were closed, and most macrophages were located near the regeneration-competent dorsal iris (Fig. 2A iv).At 15 dpl, once the lens vesicle had formed, macrophages were located in the aqueous chamber (Fig. 2A v).By 30 dpl, very few macrophages were detected in the anterior eye chamber (Fig. 2A vi).Our data suggest that the macrophage mediated innate immune response to lentectomy in newts is resolved after regeneration, through which increased number of macrophages transiently populate the different anatomical structures of the eye.
Lens removal triggers a complex early response with a strong immune signature Wound healing involves distinct phases that are well characterized at the molecular level in mammals.To characterize this highly complex response in newts, especially with regard to lens regeneration, we performed bulk RNA sequencing on two newt species (Pleurodeles waltl and Notophthalmus viridescens).We investigated dorsal irises from uninjured eyes, as well as irises following lentectomy at early time points corresponding to wound healing phases of lens regeneration.Under homeostatic conditions, the vertebrate eye is considered to be an immune-privileged organ due to several structural features and immunomodulator mechanisms that act together to limit in ammation (80, 81, 82).As anticipated, prior to lens removal, we observed relatively low expression of immune-related transcripts in both newt species (Fig. 2B&C).Lentectomy triggered a complex immune response, indicated by the upregulation of transcripts involved in the activation of signaling pathways pertaining to in ammation, ECM remodeling, pattern recognition, macrophages/monocytes, vascular development, complement activation and angiogenesis (Fig. 2B&C).Consistent with the macrophage dynamics we observed from the mpeg1:eGFP line, RNA sequencing revealed a marked upregulation of macrophage related transcripts at 4 dpl (Fig. 2A-C).A time course differential expression analysis in Notophthalmus identi ed time-dependent regulation of homologs associated with the KEGG pathways: TNF Signaling Pathway, Toll-Like Receptor Signaling Pathway, In ammatory Mediator Regulation of TRP Channels, and ECM-Receptor Interaction (Supplementary Figure S2).We observed the signi cant up-regulation of several transcripts exhibiting homology to well-known pro-in ammatory cytokines such as interleukin-1 beta (IL-1β) and tumor necrosis factor alpha (TNF-a), as well as anti-in ammatory cytokines such as interleukin-10 (IL-10) between 6 and 24 hpl in both species (Fig. 2B&C, Supplementary Figure S2 A&C).These observations are consistent with previous reports that show an early upregulation of anti and pro-in ammatory transcripts in regeneration competent species (12,15,16).By 4dpl, the expression of these cytokines was reduced in Notophthalmus viridescens, and transcripts encoding proteins that can function as in ammatory mediators such as CD200R, AIF1, ALOX5AP, ALOXE3 and PPARG were upregulated in both species.
Transcripts encoding protein products resembling ECM components, such as collagens (COL3A1), bronectin (FN1), and hyaluronic acid (HAS2), were also upregulated following lens removal (Fig. 2B&C, Supplementary Figure S2D).In addition, we observed the upregulation of transcripts that exhibited high homology to effectors of ECM remodeling, such as MMPs, ADAMs, and TIMP1 (Fig. 2B&C).Our bulk RNA sequencing data paves the way to a better understanding of a genetic response to lentectomy that triggers lens regeneration in newts.

Macrophage depletion inhibits regeneration by affecting cell cycle re-entry of iPECs
To test the role of macrophages during the early stages of lens regeneration, we injected control liposomes or clodronate liposomes into the eyes of the two newt species (Fig. 3A).Liposomes are phagocytized by macrophages and, if carrying clodronate, ultimately induce apoptosis (83, 84).Lens regeneration was evident by 30 dpl in control treated animals, but not in macrophage-depleted eyes, as indicated by the absence of a lens and A-Crystallin (a lens-speci c marker) in both species (Fig. 3B, C).In addition, histological assessment revealed several morphological and cellular abnormalities in the eye and near the dorsal and ventral aspects of the iris.Furthermore, an unusual cellular accumulation in the anterior and posterior eye chamber was observed, resembling the formation of scar-like tissue (Fig. 3B,   C).The inhibition of lens regeneration, morphological alterations, and cellular accumulation phenotypes were observed in 100% of the cases tested (n = 40 per species).These data indicate that macrophages are essential to achieve lens regeneration in newts, as their depletion leads to the formation of scar-like tissue instead of the formation of a new lens.
Next, we sought to explore potential mechanisms by which macrophage depletion could inhibit lens regeneration.Previous work suggested that macrophage depletion affects the survival of progenitor cells during n regeneration in zebra sh (85).We tested if similar mechanisms take place during newt lens regeneration by TUNEL staining.We found that macrophage depletion using clodronate liposomes did not lead to apoptosis of iPECs in the early stages of lens regeneration (up to 10 dpl; Supplementary Figure S3).As reported previously, apoptotic nuclei were observed at basal levels in transdifferentiating lens epithelial cells (LECs) in control-treated animals once a lens vesicle is formed (86).
One of the most prominent events after lentectomy is the cell cycle re-entry of terminally differentiated iPECs (37).We tested whether cell cycle re-entry is affected upon macrophage depletion by analyzing EdU incorporation in the lentectomized irises in Pleurodeles waltl.We observed that macrophage depletion changed the dynamics of cell cycle re-entry of dorsal iPECs (Fig. 3D).Based on a Negative Binomial Regression analysis, we found strong evidence of differences (p < < 0.05) in the log ratio of mean EdU + cells to HOECHST + cells at 4 dpl (Fig. 3E).Note that we provide a justi cation in the Methodology section for use of Negative Binomial regression instead of a t-test; the t-test based analysis also yields p < 0.05.In control eyes, cells located inside the lens vesicle and lens epithelium were EdU + at 10 dpl and 15 dpl respectively, whereas in clodronate liposome treated eyes, a lens vesicle failed to form, and EdU + cells were detected inside the iris and inside the vitreous and aqueous chambers.We noticed that the regenerating lens appears bigger and more developed at 15 dpl in control treated compared to the mpeg1:GFP animals at the same time (compare PBS-liposome group, 15dpl in Fig. 3D with 15dpl in Fig. 2A).This observation is consistent with our previous study showing that lens regeneration is delayed in older Pleurodeles waltl (36).Our data support that, while macrophage depletion does not cause an apoptotic response, macrophages may play a role in the induction of cell cycle re-entry of iPECs.

Macrophage inhibition prolongs in ammation, alters ECM remodeling, and causes a brotic-like response
To characterize the effects of macrophage depletion on the ensuing in ammatory response after lentectomy, we performed RT-qPCR to measure changes in the expression of the pro-in ammatory cytokine, IL-1β.Consistent with our bulk RNAseq experiments (Fig. 2B-C), IL-1β expression was upregulated after lentectomy and returned to basal levels by 30dpl in PBS liposome-treated eyes (Fig. 4A).However, in clodronate-treated eyes, IL-1β expression was signi cantly higher than controls at 10 dpl and remained higher than controls even at 30 dpl (Fig. 4A).
We then used spectral domain optical coherence tomography (SD-OCT) to image the morphological changes that accompany macrophage depletion.The non-invasive nature of SD-OCT allows us to monitor the lens regeneration process from the same newt in real time (70).Using SD-OCT, we have recently shown that ECM remodeling is highly correlated with lens regeneration (36).At 4 dpl, cloudy opacity was observed in the anterior eye chamber between the dorsal and ventral iris of both PBS and clodronate treated eyes, indicating an accumulation of ECM (Fig. 4B, Video 4).By 10 dpl, the opacity was cleared from the anterior chamber of control eyes and a newly formed lens vesicle was observed at the dorsal iris by 21 dpl (Fig. 4B, Fig. 4B-Video 5 & 6).In contrast, ECM failed to clear and progressively worsened from macrophage-depleted eyes (Fig. 4B, Fig. 4B-Video 7, 8 & 9).Several other morphological abnormalities were detected in the newt eye by 21 dpl.The aqueous humor that lls the area between the cornea and iris failed to reform in clodronate-treated eyes, and bright spots were observed in the vitreous and around the dorsal and ventral irises (Fig. 4B).We interpreted the bright spots as the cellular accumulation shown in Fig. 3B-C.To con rm our interpretation of the SD-OCT ndings, we used histology and picrosirius red staining to detect collagen bers, a major component of the ECM (36).Collagen accumulation appeared to be larger and denser in the anterior chamber of clodronate treated eyes compared to control eyes at 4 dpl (Fig. 4C).In agreement with our SD-OCT interpretations, we observed that collagen bers had begun to clear out at 10 dpl in control eyes and, by 30 dpl, very little staining was detected in the eye chambers.In contrast, collagen staining remained at 30 dpl in clodronate-treated eyes (Fig. 4C).Furthermore, we found that at least some of these cells are myo broblasts, a cell type involved in ECM remodeling, immune modulation and angiogenesis (87) (Fig. 4D).To evaluate the kinetics of myo broblast activation and accumulation in the anterior eye chamber, we performed immunostaining with alpha smooth muscle actin ( -SMA) at 4, 10, and 30 dpl in Pleurodeles waltl.While at 4 dpl we detected very few -SMA + cells in both experimental conditions, by 10 dpl, the accumulation of -SMA + cells was evident only in the anterior chamber of macrophage-depleted eyes, and this phenomenon was exacerbated by 30 dpl.In contrast, newts treated with PBS liposomes displayed substantially lower -SMA reactivity at 30 dpl relative to clodronate treatment (Fig. 4D).Collectively, SD-OCT and histology demonstrated signi cant abnormalities following macrophage depletion in the newt eye including cellular accumulation and lack of ECM clearance, which resulted in the formation of scar-like tissue at the expense of lens regeneration.

Exogenous FGF can rescue lens regeneration processes in macrophage-depleted newts
Previous studies have shown that FGF signaling pathway plays an important role during lens regeneration (88, 89, 90, 91, 92).Since macrophages have been shown to directly secrete FGF in other contexts (93,94), we hypothesized that macrophage depletion could affect the percentage of iPECs reentering the cycle through a decrease in the FGF levels in the newt eye.To test this hypothesis, we supplied exogenous FGF2 into the newt eye right after lentectomy, followed by clodronate treatment (Fig. 5A).We observed that FGF treatment caused a signi cant increase (p < 0.001) in the overall number of EdU + iPECs at 4 dpl (Fig. 5B).Astonishingly, by 30 dpl, a lens vesicle was observed in one third of the cases, stemming from the dorsal iris of FGF2-supplemented eyes following clodronate treatment (4/ 12 eyes) (Fig. 5C).In addition to rescuing regeneration, clodronate and FGF2 co-administration caused a reduction in the generalized cellular accumulation, as indicated by the lack of nuclear staining compared to clodronate and PBS co-administration (Fig. 5C).Our data show that exogenous administration of FGF2 in macrophage-depleted newts can rescue lens regeneration as well as decrease the accumulation of cellular components of the scar-like tissue.
We next evaluated the gene expression levels of several in ammatory and cell cycle-related genes via RT-qPCR analysis at 4 and 10 dpl.The expression levels of cell cycle associated gene P53 was found to be signi cantly upregulated in clodronate/FGF2 treated eyes compared to clodronate/PBS treated eyes at 10 dpl.In addition, we observed moderate evidence (adjusted p < 0.1) that E2F1 and CDK2 were upregulated at 4 dpl and 10 dpl respectively in the FGF2-treated eyes.Furthermore, the expression of the suppressor of cytokine signaling 3 (SOCS3) gene was signi cantly downregulated at 4 dpl in clodronate/FGF2-treated eyes.In contrast, we did not nd evidence for changes in the expression patterns of macrophage speci c receptor CSF1R, in ammatory agents COX2, IL-1β, TGFβ2, and TGFβ3, (Fig. 5D).We then tested if supplementing FGF in macrophage depleted eyes affected the apoptotic levels at 30 dpl.We found that apoptotic cells were still evident in the aqueous chamber near the regenerating lens and inside the cornea (Fig. 5E).Our results suggest that FGF plays an important role during iPECs cell cycle re-entry and supplementation of exogenous FGF is su cient to start the regeneration process even in the absence of macrophages.

Late administration of clodronate enhances apoptosis and pro-in ammatory signals
To examine if newt macrophages play a role during the later stages of lens regeneration after the critical window of lens vesicle formation is passed, we treated eyes with clodronate or PBS liposomes at 10, 12, and 14 dpl (Fig. 6A).Even though a lens was detected by 30 dpl, and LECs were positive for EdU and phospho-histone H3 (PHH3), the lens appeared smaller in all clodronate-treated eyes (Fig. 6B).In addition to abnormal lens morphology (smaller size, presence of cells without ber characteristics in the lens cortex and a multilayer lens epithelium), severe cellular accumulation was observed between the lens and the ventral iris (Fig. 6B, C).Collagen ber staining revealed that ECM accumulation was evident in the vitreous and aqueous chambers following clodronate treatments (Fig. 6C).Furthermore, TUNEL + nuclei were observed inside the lens bers and near the ventral iris in clodronate treated eyes, indicating that macrophages may play a role in preventing cell death by apoptosis in the late phases of the regenerating lens (Fig. 6D).Similar observations were noted when clodronate administration was initiated at 20 dpl (Supplementary gure S5).To complement, we tested for the expression of selected immune and in ammation related targets via RT-qPCR and found that late clodronate administration caused an increase in expression of the pro-in ammatory cytokine IL-1β and the macrophage-speci c receptor CSF1R at 30 dpl (Fig. 6E).Expression patterns of FGF2, TGFB2, TGFB3, MMP3, and MMP9 were not signi cantly changed (Fig. 6E).Altogether, these data suggest that macrophages are also necessary for the proper regeneration of the lens in the later phases of the process, potentially controlling survival of lens cells and resolving pro-in ammatory signals.

A new injury helps to resolve the brotic phenotype and can re-activate lens regeneration
The results observed above led us to believe that macrophages in newts were playing an anti-brotic role during the injury response.This made us question whether macrophages, if given su cient time to recover in number following clodronate depletion and lentectomy, could resolve the brotic injury that was triggered by their absence.To test this, we treated eyes with clodronate or PBS liposomes at 0, 2, and 4 dpl and monitored the animals using SD-OCT for 100 days, as well as H&E, EdU staining, and A-Crystallin immunohistochemistry, and we found no evidence of proliferation or lens formation at 100 dpl in all cases that were treated with clodronate (n = 10) (Fig. 7A).Startlingly, we also observed a microphthalmic phenotype in 5/10 eyes that were treated with clodronate (Fig. 7B).
Since we observed a sustained in ammatory response with macrophage depletion, we hypothesized that the macrophages returning to the eye chamber were met with such an in ammatory microenvironment, that the macrophage phenotype was immediately polarized to a damaging, and hence continued, proin ammatory response upon entry into the eye.We wanted to test if a secondary injury was su cient to reprogram the macrophages in order to resolve the chronic in ammation and brosis.To do that, we surgically removed a piece of the dorsal iris (iridectomy) at 60 dpl in PBS and clodronate-treated eyes without removing the brotic mass that was present in the anterior eye chamber (Fig. 7C).It has previously been shown that following iridectomy, the iPECs proliferate to replace the missing tissue, and a lens is formed by the transdifferentiation of the newly formed iPECs (95).Similar to these observations, we detected a newly formed lens following iridectomy in PBS-treated eyes (Fig. 7C).Interestingly, we found that in 3/8 cases where iridectomy was performed in clodronate-treated eyes, regeneration was initiated, indicated by the presence of an A-Crystallin + lens vesicle at the dorsal iris (Fig. 7C).Importantly, the cellular accumulation and brotic phenotype was resolved from the anterior chamber in the cases that lens regeneration was induced (Fig. 7C).

Discussion
The requirement of macrophages for successful regeneration has been demonstrated in a wide range of vertebrates including axolotl, frogs, zebra sh, lizards, and African spiny mouse, but not in newts, the champions of regeneration, and never in the lens (11,12,15,16,17,18,19,20,40,96).Here we used a newly generated transgenic newt line, mpeg1:GFP to show the dynamic recruitment of macrophages following lentectomy.Using bulk RNAseq data from two newt species, we uncovered upregulated expression of genes linked to macrophages/monocytes, in ammation, and other processes.We discovered that macrophage depletion during the early stages after lens removal inhibited regeneration in both Notophthalmus viridescens and Pleurodeles waltl.We went on to show that early macrophage depletion signi cantly reduced iPEC proliferation, but did not affect apoptosis.When we supplemented with exogenous FGF2, iPEC proliferation, the in ammatory pro le, and regeneration were rescued.Furthermore, we showed that macrophage ablation failed to resolve the normal in ammation and ECM accumulation observed with injury, resulting in a deleterious eye environment which led to unresolved cellular accumulation, brosis, and eventually leading to microphthalmia.While a pro-in ammatory signature was detected when macrophages were depleted at later stages of regeneration, this experimental paradigm showed no changes in proliferation but presence of apoptosis.Importantly, we showed that re-initiation of an injury in the eye, broke the progressive cycle of in ammation, resolved the chronic brosis, and re-started the regenerative process.Collectively our data indicate that macrophages stand at a critical turning point during newt lens regeneration and determine the balance between scarring and regenerative phenotypes (Fig. 8).
Despite being renowned for their regeneration capabilities, the utilization of newt species such as Notophthalmus viridescens and Cynops pyrrhogaster as model organisms has proven a di cult task (97).Their gigantic genomes and complex life cycles have hampered the creation of molecular tools and the development of genetically modi ed animals.However, the application of recent technologies to Pleurodeles waltl, a newt species with similar regenerative capabilities (36, 97) which is easy to breed in the laboratory with a shorter reproductive cycle, and a recently sequenced genome, removed these roadblocks and brought newts into the molecular era (45,59,97,98,99,100,101,102,103).Here, we utilized Pleurodeles waltl to develop the rst macrophage reporter line in newts, using the macrophageexpressed gene 1 (mpeg1) promoter to drive the expression of eGFP (38,74).We validated the mpeg1:eGFP transgenic line by testing the colocalization of eGFP + cells with known macrophage markers (F4/80, L-plastin, CSF1R) and demonstrating their phagocytic activity through local incorporation of GeRPs in several different tissues.We used this line to evaluate the spatiotemporal kinetic accumulation of macrophages in the newt eye during lens regeneration.Consistent with previous publications that used electron microscopy to characterize macrophage involvement during lens regeneration, we found a clear presence of eGFP + cells between 4-10 dpl (34).Interestingly, most of eGFP + cells were near the regeneration-competent dorsal iris.The signi cance of this nding has not been evaluated in this report.However, future studies focusing on how the dorsal/ventral microenvironment affects the recruitment and phenotype of macrophages could be instrumental towards our understanding on the role of macrophages during scar-free healing.Nevertheless, the introduction of the mpeg1:eGFP transgenic line is a powerful addition to the regeneration toolbox and opens the door for a wide range of applications that were not before possible in newts.
It is hypothesized that the transition from wound healing to regeneration or scar formation is controlled by the complex interactions between immune cells, damaged tissues, and surrounding microenvironment (9,104).It is likely that the interaction between macrophages and iPECs during the early steps of wound healing provide critical signals for the initiation of cell cycle re-entry, dedifferentiation and transdifferentiation, as well as the resolution or propagation of in ammatory signals.We sought out to investigate this hypothesis and delineate the signaling pathways that occur during the early stages of wound healing in the newt eye via RNA sequencing.Our results showed an early upregulation of several genes associated with the resolution of in ammation, which is reminiscent to expression pro les reported in other regeneration-competent tissues, in contrast to a prolonged expression of pro-in ammatory signals that are dominantly observed in non-regenerative animals (12,104,105,106,107).In addition, we detected the upregulation of several matrix remodeling genes, including several members of the MMP family.The importance of MMP enzymes to degrade ECM and promote regeneration has been demonstrated in several species and tissues (18, 108, 109).Furthermore, our data indicates the up regulation of genes in the lipoxygenase family.Lipid mediators play a key role in determining the magnitude and duration of in ammation in several species, however the role of these enzymes in regenerative animals is not yet well understood (110).Overall, our RNA sequencing data suggest that during the early stages of lens regeneration, immune cells function to promote mechanisms that favor in ammation resolution and matrix remodeling and create a regeneration permissive environment.Further studies are required to speci cally characterize the polarization state of macrophages and de ne the kinetics of M1 vs M2 class switching during lens regeneration in newts.
Macrophages can directly or indirectly regulate several aspects of the wound healing and regeneration processes.In zebra sh, for example, it was shown that macrophage depletion affects the survival of progenitor cells during n regeneration (85,111).We tested if this is true during lens regeneration in newts and found no evidence of apoptosis in the iPECs of the dorsal iris at the early stages of regeneration in the absence of macrophages.We did, however, observe changes in the rate of cell cycle re-entry and proliferation of iPECs.Since macrophages can support proliferation through secretion of growth factors (112,113), we tested if the addition of growth factors will counteract the effect of macrophage depletion in the newt eye.FGFs represent good candidates due to their known role in lens regeneration (88,89,90,91).Astonishingly, when we exogenously supplemented a source of FGF in the absence of macrophages, the cellular accumulation was resolved and iPEC proliferation and regeneration were recovered.In addition, we showed that expression of SOCS3 was downregulated in eyes that were supplemented with FGF2.Inhibition of SOCS3 has been shown to promote liver and axon regeneration in mammals (114,115).How macrophages either directly or indirectly affect the expression of FGFs in the newt eye requires further exploration.One possible explanation is that macrophages directly secrete FGF into the iris to promote proliferation (94,112).It's also possible that macrophage depletion indirectly affected FGF levels by preventing its tra cking from nearby tissue sources.It has been documented, that during lens regeneration, the neuroretina secretes growth factors that are necessary for the reprogramming of iPECs (116, 117).We showed that lentectomized eyes treated with clodronate liposomes exhibited higher levels of collagen in the aqueous and vitreous cavities.Therefore, it is possible that excessive amounts of ECM accumulation in the eye cavity could negatively affect the tra cking of growth factors from the neuroretina into the iPECs (118,119).
Several studies have demonstrated that initiation of in ammation is necessary for successful regeneration (85, 120,121,122,123,124).However, the magnitude and duration of in ammation is also key for determining the wound healing outcome (125,126).It is hypothesized that resolution of in ammation occurs much faster in regeneration-competent animals compared to non-regenerative species such as mammals (127,128).In fact, pharmacological attenuation of in ammation promoted tissue repair in regeneration-incompetent animals, thus demonstrating the importance of resolving in ammation during the early stages of wound healing (12,129).Increasing evidence suggests that macrophages are responsible for placing the necessary restraints in the in ammatory reaction during the early stages of regeneration (15,16,85,130).Consistent with these studies, we show that macrophage depletion in the newt eye prolonged the increased expression of the pro-in ammatory cytokine IL1-β.
In mammals, prolonged expression of in ammatory genes often leads to brosis and ultimately scarring (131,132,133).This phenomenon is exacerbated during aging and repeated damage (134,135).Salamanders, on the other hand, have mastered scar-free healing and are thought to be resilient to brosis.This ability was highlighted in a pioneering study, where the lens was removed 18 times from the same newts in a period of 19 years, and each time, the lens was perfectly regenerated without any signs of brosis or scarring (136,137).It is tempting to speculate that the resistance to brosis in these animals could be due to the fact that newt macrophages adopt an earlier and more robust antiin ammatory phenotype relative to their mammalian counterparts.It is also possible that newt macrophages can directly secrete collagens, MMPs, or growth factors that would modulate myo broblast activation and shape extracellular matrix remodeling (18, 94,138,139,140).We tested the kinetics of ECM remodeling during normal lens regeneration and after early macrophage depletion with clodronate.
Using our SD-OCT imaging modality in combination with histology, we detected the unprecedented formation of scar-like tissue in the newt eye following lentectomy with macrophage depletion and that ECM failed to clear.In addition, collagen composition and myo broblast activation was altered with macrophage depletion.Our ndings suggest that under normal conditions, newt macrophages play a critical role in preventing myo broblast accumulation and ne-tune matrix remodeling.De ning the composition of ECM during injury in newts and identifying the complex interaction between ECM and immune cells that lead to its breakdown will be bene cial towards our understanding of ECM involvement during regeneration.
Interestingly, depletion of macrophages during late stages of regeneration, after the lens vesicle has formed, did not affect proliferation but it increased apoptosis in the lens bers and around the ventral iris.
Clodronate treatment also arrested the regeneration process, induced expression of the pro-in ammatory cytokine IL1β, and resulted in unresolved cellular accumulation.Our ndings suggest that macrophages also play important roles after the critical window of transdifferentiation: they contribute to promote and maintain homeostasis in the eye by reducing the pro-in ammatory environment, preventing scar formation and keeping apoptosis at bay.
Surprisingly, when we monitored clodronate-early-treated eyes for 100 dpl, we found that the scar-like tissue components (cellular in ammation, ECM accumulation, and brosis) progressively worsened and, in most cases, microphthalmia was observed.These observations are interesting because macrophage depletion with clodronate is transient, yet the return of macrophages appeared unable to resolve the ECM accumulation that occurred in the absence of macrophages at the key early timepoints.It is unclear whether the macrophages returning to the brotic lesion and altered microenvironment have lost the ability to regulate ECM dynamics or whether the ECM has progressed into a state that the returning macrophages are unable to modify.Astonishingly, introduction of a secondary injury without removing the scar tissue, by removing a small piece of the dorsal iris was su cient to resolve the deleterious effects of macrophage depletion and re-start the regeneration process.These ndings are noteworthy and suggest that signaling events are capable of reprogramming newt macrophages into a pro-resolving and anti-brotic phenotype, even in the presence of a brotic environment.Eliminating myo broblasts has been a long-standing goal for developing therapies to treat brosis in humans.In newts, however, macrophages can achieve this when activated in response to a new injury.Even more impressively, they can do this after chronic in ammation and at a point which stable brotic disease has already been established (138).Consistent with our observations, Godwin et al., have previously shown that reamputation of brotic limbs after macrophage depletion in axolotls resulted in regeneration rescue (16).
However, in these experiments the entire brotic environment was removed by amputation, whereas in our study, we kept the vitreous and aqueous chambers (area lled with collagens, cellular accumulation and myo broblast) and only removed the source of regeneration (dorsal iris epithelium) accentuating the signi cance of our ndings.Our ndings illustrate that newts can not only re-start a previously hampered regeneration process, but they are able to correct previously established brotic tissue when exposed to a new insult.Future studies comparing the phenotype of newt macrophages during normal lens regeneration in control animals with those recruited to the brotic environment after clodronate treatment and those after secondary injury has potential therapeutic interest, as this approach could give us clues to which polarizing factors are responsible for inducing anti-brotic and pro-regenerative outcomes in newts.
There are several limitations in the present study.First, bulk transcriptomic methods were used to characterize the molecular mechanisms that promote wound healing and in ammation resolution during the early stages of regeneration.The iris epithelium and stroma consist of a heterogeneous cell population (iPECs, keratinocytes, melanocytes, muscle, blood, and in ltrating immune cells such as macrophages).In axolotls and zebra sh, it was shown that not only immune cells, but also other cell types such as blastema progenitors and n epithelial cells (in limb and n respectively), can also produce cytokines that modulate the in ammatory response and in uence the regeneration outcome (85,141).
Future studies utilizing technologies with the ability to resolve cell types, such as cell sorting, single-cell RNA sequencing, and/or spatial transcriptomics will shed light on how each cell population contributes differently during lens regeneration.Furthermore, these technologies will allow us to identify the heterogeneity of the cellular accumulation that we observed following clodronate treatment.Another limitation comes from the use of liposomes to deplete macrophages.Since liposomes cannot penetrate the blood-brain barrier, they must be injected intraocularly, thus creating an additional injury to the eye.Pharmacological studies using small molecules to directly target macrophages and the generation of animals with genetically depleted macrophages will aid in further exploration of the function of macrophages during scar-free healing in newts.

Conclusions
Collectively, our results suggest that lens regeneration in newts can be added to the ever-growing list of cases that demonstrate the necessity of macrophages for successful regeneration.Identifying the common mechanisms by which macrophages contribute to scar-free healing across different regenerative species and tissues will be critical towards our goal to integrate these ndings for clinical translation and the development of therapeutic strategies.This study offers a unique perspective and a rst glimpse into the functions of macrophages to achieve successful regeneration in the newt eye: by promoting early cell cycle re-entry in iPECs, resolve pro-in ammatory signals, and prevent brotic scar formation.Furthermore, we showed that macrophage depletion during lens regeneration establishes a progressive brotic disease state in the newt eye that could lead to microphthalmia.While we showed the brotic disease is unable to resolve on its own, a secondary injury performed months later in the dorsal iris resulted in the reabsorption of the brotic scar within the eye and the initiation of lens regeneration.

Figures
Figures

Figure 4 Early
Figure 4

Figure 7 A
Figure 7 The remarkable reversal of brosis with re-injury occurred in the presence of returning macrophages, highlighting their potential role in clearing the previously established brotic disease.Taking into consideration that brosis in humans is often considered irreversible, these observations are signi cant and bear great translational potential.Our ndings establish a new experimental context in which the mechanisms behind scar-free healing, regeneration and scar reabsorption can be studied further.All procedures involving animals were performed according to according to the guidelines approved by both the Animal Care and Use Committee at Miami University, Swedish and European regulations.Miami University: (1) Title of the approved project: Investigating lens and retina regeneration in newts and axolotls; (2) Name of the institutional approval committee: Institutional Animal Care and Use Committee