Introduction

Crystalline silica (CS) is a typical inorganic particle in natural and industrial settings. Exposure to respirable CS leads to pneumoconiosis, characterized by chronic inflammation and progressive pulmonary fibrosis1,2. Though redoubled efforts were made to minimize CS exposure, stubbornly high morbidity and mortality of silicosis emphasized the hazardous burden of CS-related diseases3,4. The new emerging industries like sandblasting denim jeans and manufacturing of artificial stone benchtops reignited the emergence of silicosis around the world5. While silicosis is a preventable disease, unfortunately, patients continue to suffer from this progressive disease2. Interventions against silicosis progression are in high demand.

The inhaled CS particles deposited in the lung interstitium trigger inflammatory cascades involving innate and adaptive immune responses6. Although some early events, such as macrophages engulfing CS particles, in silicosis are clear, the following steps leading to fibrosis are less well-understood7,8,9. Different from the simple exposure-response relationship, adaptive immune response characterized as disorders of T lymphocytes orchestrate chronic inflammation and fibrogenesis10. CD4+ helper T (Th) cells have been identified as vital players in fibrotic disorders, including silicosis11. The CD4+ Th cells can be divided into pathogenic effector T cells (including Th1, Th2, and Th17 cells) and immunosuppressive regulatory T cells (Tregs), whose fate was governed by transcriptional factors T-bet, GATA-3, ROR-γt, and FOXP3, respectively12. Notably, T cell-mediated adaptive immune response is characterized as immunological memory. Memory T cells (TM) expressed memory-T-cell-associated molecule, CD44, and can be divided into several subsets13. Central memory T cells (TCM) patrol the blood and secondary lymphoid organs, while effector memory T cells (TEM) express homing molecules, allowing them access to peripheral tissues14,15. The TCM and TEM cells are abundant in circulation, while tissue-resident memory T cells (TRM) preferentially localize in barrier tissues such as the lung16,17. TRM cells rapidly respond to the invading pathogen within peripheral tissues, providing first-time and robust protection, while in some chronic inflammatory diseases, TRM cells exert pathogenic roles18. Th cell’s function in silicosis was explored19. However, these researches were largely based on the evidence of peripheral T cells. The knowledge about the contribution of lung resident TRM cells to silicosis is limited.

T cells migrate to the damaged tissue, mature, and maintain in non-lymphoid tissues, exerting enhanced effector functions compared to their lymphoid tissue counterparts. They rapidly respond to the invading pathogen within peripheral tissues, providing first-time and robust protections upon cognate antigen stimulation20. Though TRM cells have primarily been described as their protective functions, particularly in the context of pathogen infections, TRM cells, especially CD4+ TRM cells, have also been reported to be pathogenic in chronic inflammatory settings16,21.

In the pathogenesis of silicosis, the inhaled CS particles can be engulfed by macrophages but cannot be cleared, resulting in the re-release of CS particles, which leads to repeatedly, locally re-initiated chronic inflammation. Under such circumstances, inflammatory cytokines (like IL-17A, IFN-γ, and TNF-α) and profibrotic cytokines (like TGF-β and PDGFα) could promote fibroblast activation, shown by increased fibrosis-related protein, collagen I and fibronectin22. Additionally, our previous study demonstrated that cytokine-producing CD4+ T cells in the silicotic lung manifested TM cell characterization23. Given the situation of repeated CS stimulation and the immunologic memory function of TRM cells, we hypothesized that CD4+ TRM cells are involved in the pathogenesis of silicosis.

To this end, we utilized a murine model of silicosis delineating the effects of CS particles on CD4+ TRM cells and further explored its pathogenic role in silicosis. Our results demonstrated CS particles induced substantial accumulations of pulmonary CD4+ TRM cells. We further proved the source and function of CD4+ TRM cells in the pathogenesis of silicosis. The CD69+CD103 CD4+ TRM subset manifested robust pro-inflammatory responses, whereas the CD69+CD103+ CD4+ TRM subset was immuno-suppressive. Significantly, targeting the maintenance and function of pathogenic lung CD4+ TRM cells exerted protective effects against silicosis.

Results

CS particles stimulated CD4+ TRM cell emergence and expansion along with silicosis progression

First, we utilized the in vivo labeling method distinguishing tissue-resident cells that are commonly used in multi-vascular tissues (Fig. 1a)23. We observed a remarkable appearance of CD4+ TRM cells (CD44+CD45 i.v.) in the silicotic lung (Fig. 1b), whereas few CD4+ TRM cells were found in saline-treated mice. Unlike the circulating CD4+ TEM cells (CD44+CD45 i.v.+), CD4+ TRM cells surged continuously with the progression of silicosis (Fig. 1c and Fig. S1b), suggesting a link between the emergence of CD4+ TRM cells and silicosis progression. Furthermore, compared with saline-treated mice pulmonary CD4+ TRM cells, the elevated expressions of cell retention markers, CD69, CD103, and CXCR6 further confirmed silicotic CD4+ TRM cells possessed lung retention ability (Fig. 1d). Furthermore, the CD69+CD103 subset was observed in both saline- or CS-treated lung, whereas the CD69+CD103+ subset was distinct in silicotic lung (Fig. 1e). With the expansion of CD4+ TRM cells, there was an increasing number of CD69+CD103+ and CD69+CD103 subsets (Fig. 1f, g), implying their pathogenic roles in silicosis. Comparatively, the phenotype of circulating CD4+ TEM cells was analogic in saline and CS-treated mice, which differed from TRM cells (Fig. S1c). Notably, CD69 and CD103 expressions on splenic CD4+ TEM cells were not affected by CS injury in the lung (Fig. S1d), indicating CS led to a tissue-local response. Collectively, these data demonstrated CS stimulated the emergence and expansion of pulmonary CD4+ TRM cells that were tightly related to silicosis progression.

Fig. 1: CS particles stimulated CD4+ TRM cell emergence and expansion along with silicosis progression.
figure 1

a Schematic showed CD45-APC-Cy7 antibody intravenous (i.v.) injection to distinguish tissue-resident or circulating leukocytes. Circulating leukocytes were labeled. b Flow cytometry (FC) analysis of pulmonary CD4+ TRM cells (CD45inject CD44+) and TEM (CD45inject+ CD44+) cells. c Percentages and counts of the CD4+ TRM cells were compared at the indicated time points. W week. d Representative FC analysis of CD4+ TRM cells for CD69, CD103, and CXCR6 expression, respectively. e FC analysis of CD4+ TRM cells for CD69 and CD103 expression. f, g The graph showed percentages and counts of CD69+ CD103+ (f) and CD69+ CD103 subsets (g) in CD4+ TRM cells of saline or CS-treated mice at the specified time points. The bar graphs are the combined results of at least three independent experiments. Individual mice are plotted on the graphs, n = 5 biologically independent animals. Values are reported as the mean ± SD. P value was determined by one-way ANOVA followed by Tukey’s test.

Pulmonary CD4+ TRM cells mediated severe lung inflammatory response to CS particles, promoting the pathogenesis of silicosis

Given the immunologic memory role of TRM cells, we next sought to explore the response of CD4+ TRM cells to CS particles by T-cell transfer studies. CD4+ T cells were sorted from the 8 weeks post-CS-treated lung containing certain TRM cells (TRM), the spleen under CS treatment only involving circulating TEM cells (TEM), or the spleen of saline-treated mice, including naive T cells (TN) to CS particles, respectively. The sorted cells were adoptively transferred into Rag1–/– mice that lack T cells, and then all Rag1–/– mice were treated with CS (Fig. 2a). H&E staining revealed that CS-induced the severest inflammatory cell recruitment and infiltration in the mice lung transferred with TRM cells. In contrast, the mice who transferred TN or TEM cells exerted relatively mild responses (Fig. 2b, c). These phenotypes were further confirmed by the transcripts of cytokines associated with adaptive immunity, including Ifng, Tnfa, and Il17a, but not Il6 (Fig. 2d). We further adopted flow analysis to the lungs. Significantly, more pulmonary resident T cells were observed in the mice transferred with CD4+ TRM than those of naive counterparts (Fig. 2e), highlighting a rapid reaction and expansion of the CD4+ TRM cells to CS particles. In supporting the notion, a higher Ki67, cell proliferating marker, was observed in the cells from the mice transferred with CD4+ TRM (Fig. 2f). While lung resident T cells in the mice transferred with CD4+ TEM cells resembled those transferred with certain CD4+ TRM cells (Fig. 2e), implying the ability of TEM cells converting into TRM cells. Additionally, we discovered the ratio of CD103 to CD103+ in CD69+ TRM was affected by the distinct cellular sources, while more CD69+CD103 subsets resided in the CD4+ TRM transferred mice (Fig. 2g). These results implied the CD69+CD103 TRM cells mediated pro-inflammatory effects to the invaded CS particles.

Fig. 2: Pulmonary CD4+ TRM cells mediated severe lung inflammatory response to CS particles.
figure 2

a The reconstitution sketch of Rag1–/– mice with different cell origins. The MACS-purified CD4+ T cells and PBS were i.v. transferred into Rag1–/– mice before CS instillation. The recipient mice were analyzed one week after the CS instillation. b H&E staining to the lung sections of distinct reconstituted Rag1–/– mice. Scale Bar = 500 μm (upper) and 200 μm (lower). c The inflammation scores were assessed in the lung sections. d Relative RNA levels of Ifng, Tnfa, Il17a, and Il6 in each group. e FC analysis of pulmonary resident CD4+ T cells in transferred Rag1–/– mice. f Flow histogram indicated Ki-67 intensity in lung-resident CD4+ T cells. g FC analysis compared the proportions of CD103 to CD103+ in the CD69+ CD4+ TRM cells. h The scheme indicated the time points of CS instillation: Single CS instillation was treated at 0 W (3 mg in 50 μL), while 50 μL sterile saline per week was instilled at 1 and 2 weeks. Repeated CS instillation was treated at 0, 1, and 2 W (1 mg in 50 μL per week), respectively. Both groups were sacrificed 10 W after the first CS treatment, with an equal amount of CS particles in total. i Masson’s trichrome staining to the lung sections of different groups. Scale bar = 500 and 200 μm. The bar graph showed the fibrosis-positive area. j Relative mRNA levels of Col1a1 and Fn in each group. k FC analysis of pulmonary CD4+ TRM cells in single or repeated CS-treated mice. l Relative mRNA levels of Il17a, Ifng, and Tnfa in each group. m FC analysis showed the ratios of CD103 to CD103+ subsets in CD69+ CD4+ TRM cells. For panels ag, n = 4–5 biologically independent animals, n number also indicated independent experimental replicates. Littermate recipient Rag1–/– mice and donor C57BL/6J mice were used in these experiments. Values are reported as mean ± SD. P value was determined by one-way ANOVA followed by Tukey’s test. For panels, hm, n = 6–7 biologically independent animals, the bar graphs are the combined results of at least three independent experiments. Individual mice are plotted on the graphs. Values are reported as mean ± SD. P value was determined by unpaired two-tailed Student’s t-test.

Considering persistent particle exposure in the working environments24,25, we adopted a repeated CS exposure model (Fig. 2h), exploring the role of CD4+ TRM cells in mediating silicosis. Though the mice were exposed to the same amount of CS particles in total, repeated exposure resulted in severe pulmonary fibrosis, which was demonstrated by Masson’s trichrome staining and elevated Col1a1 and Fn transcripts (Fig. 2i, j). We further did flow analysis to the CS-exposed lung, demonstrating that the repeated CS exposure surged more CD4+ TRM cells in the lung than the one-time challenge (Fig. 2k). The transcripts of proinflammatory cytokines Il17a, Ifng, and Tnfa were also higher in the CS-repeated exposed lung (Fig. 2l). Precisely, we dissected that there were more CD69+CD103 subsets in the CD4+ TRM cells (Fig. 2m), which implied repeated CS stimulated specific TRM subsets expansion. In synthesis, these results demonstrated that CD4+ TRM cells exerted immuno-memory to the CS particles mediated the pathogenesis of silicosis, and the CD69+CD103 TRM subsets possessed robust pathogenic capacity to silicosis.

CS-induced TRM cells derived from circulating T cell recruitment and proliferation in situ

We next sought to study the origin of CD4+ TRM cells in silicotic lungs. By applying FTY720 treatment in C57BL/6J mice, we blocked leukocyte egress from the peripheral lymphoid tissue, minimizing the recruitment of circulating leukocytes26. Particularly, the treatments at different time points within silicosis progression were employed to elucidate the origin and maintenance of CD4+ TRM cells (Fig. 3a). H&E staining demonstrated that FTY720 treatment exerted protective effects on silicosis. However, though half-time intervention (4–8 weeks) reduced inflammatory cell recruitment, we could not get an equal attenuated phenotype compared to full-time blockage (0–8 weeks) (Fig. 3b, c), suggesting that peripheral circulating cells would turn into TRM cells. Full-time and half-time FTY720 treatment resulted in a significant reduction of circulating CD4+ T cells shown by in vivo labeling. On the contrary, the existence of TRM cells implied that CS-induced pulmonary TRM cells actively expanded in situ (Fig. 3d). Specifically, full-time and half-time FTY720 treatment resulted in an equal vanish of CD4+ TRM cells in number, indicating that the recruitment after 4 weeks post-CS-treatment was essential for CD4+ TRM cells. Accordingly, high levels of Ki-67 were observed in the TRM cells but not impaired by FTY720 treatments (Fig. 3e). Notably, we noticed affected ratios of CD103 to CD103+ in CD69+ TRM by the FTY720 intervention that full-time FTY720 treatment resulted in a high portion of CD69+CD103 TRM subsets (Fig. 3f) that analog to the phenotypes of previous transfer experiments. We next explored the proportion of Tregs in the TRM cells and got a lower ratio in the full-time FTY720-treated mice lung (Fig. 3g), reminding us that TRM-Tregs were more dependent on the replenishing of peripheral circulating cells.

Fig. 3: CS-induced TRM cells derived from circulating lymphocyte recruitment and in situ proliferation.
figure 3

a The scheme indicated the time points of FTY720 treatment (20 μg in vehicle per time). FTY720 treatment from 4 weeks, in which circumstance, lymphocytes could be recruited into the lung at the inflammatory stage but blocked at the fibrogenesis stage. b H&E staining to the lung section of different treated mice. Scale Bar = 500 μm (upper) and 200 μm (lower). c The inflammation scores were assessed in the lung sections. d FC analysis of CD4+ TRM cells in the lungs of distinct FTY720-challenge. The bar graph shows percentages and counts of CD4+ TRM cells. e Flow histogram indicated Ki-67 expression in CD4+ TRM cells. f FC analysis of CD4+ TRM cells for CD69 and CD103 expressions. The bar graph displayed the ratios of CD103 to CD103+ in the CD69+ TRM cells. g FC analysis of Tregs (FOXP3+) in the CD4+ TRM cells. The bar graph shows the percentages of TRM-Tregs. h Schematic overview of parabiosis experiment. CD45.1/1 and CD45.1/2 mice were approximated with sutures. With 14 days’ recovery, CD45.1/1 mice were treated by CS particle instillation and analyzed 7 days later. i Typical FC plot of circulating blood leukocytes of the parabiont. j Flow plot analysis of the CD4+ TRM cells and circulating TEM cells of the CS-treated conjoined mice. CD45.2+ TRM cells indicated the recruited cells from circulating. k Typical flow plots showed CD69 and CD103 expressing patterns on TRM-Teff cells or TRM-Tregs. Flow plot analysis of CD45.2+ cells in CD69+CD103+ Tregs, CD69+CD103+ Teffs, CD69+CD103- Tregs, and CD69+CD103- Teffs in CD4+ TRM cells. For panels ag, n = 5 biologically independent animals, n number also indicated independent experimental replicates. Individual mice are plotted on the graphs. Values are reported as the mean ± SD. P value was determined using one-way ANOVA followed by Tukey’s test. For panels j and k, n = 4 biologically independent animals, n number also indicated independent experimental replicates. CD45.1/1 and CD45.1/2 mice approximated with sutures were used in these experiments. Values are reported as the mean ± SD. P value was determined by paired two-tailed Student’s t-test.

We further performed a parabiosis surgery to illuminate the source of the TRM cells in the CS-injured lung26,27. To do this, naive CD45.1/2 mice were cojoined with naive congenic mice (CD45.1/1) (Fig. 3h). The blood circulation between parabionts was established 14 days after the surgery, indicated by an equal portion of CD45.1 and CD45.2 lymphocytes in the blood (Fig. 3i), after which the CD45.1/1 congenic mice were exposed to CS particles. 7 days later, we checked the component of circulating TEM cells and the pulmonary TRM in the CS-exposed mice. We found the composition of CD45.2+ lymphocytes in circulating TEM cells was equal to those in the blood (Fig. 3j). Notably, we found that there were CD45.2+ cells emergence in the CS-treated CD45.1/1 mice pulmonary TRM cells (Fig. 3j), suggesting circulating T cells contributed to TRM cells. Moreover, we found both the TRM-Teff cells and TRM-Tregs were dependent on circulating T cells, manifested by a portion of CD45.2+ T cells in the congenic mice (Fig. 3k). We also observed that TRM-Tregs expressed higher CD103 than the TRM-Teff cells (Fig. 3k), which reminded us that the adhesive molecule CD103 was related to distinguish different CD4+ TRM subsets. However, when we scrutinized TRM-Tregs, we found the highest portion of CD45.2+ cells in the CD69+CD103+ TRM-Tregs, proving that CD103+ TRM-Tregs are dependent more on circulating cell replenishment (Fig. 3k). Collectively, these data demonstrated that CS-induced pulmonary CD4+ TRM cells came in two ways: recruited from circulation and proliferating in situ. The TRM-Tregs were dependent more on replenishment from circulating cells.

Differential CD69 and CD103 expressing patterns defined silicotic CD4+ TRM cells into relatively functional distinct lineages

Previous results suggested that adhesion molecule expressions may be related to the constitution of TRM subsets. Next, we aimed to scrutinize the phenotype of CD4+ TRM cells in silicotic lungs. The CD4+ TRM cells were divided into 4 subsets on differential CD69 and CD103 expressions28 (Fig. 4a). Interestingly, the CD69+CD103 and CD69CD103 subpopulations expressed higher T-bet, indicating pro-inflammatory Th1 cells were enriched (Fig. 4b). The comparison of ROR-γt manifested that the CD69+CD103 subpopulation expressed the highest level (Fig. 4c). The CD69+CD103+ subset exerted the highest GATA-3 expression (Fig. 4d). A higher portion of FOXP3+ was observed in CD69+CD103+ and CD69CD103+ subsets (Fig. 4e). Cell retention markers could depict TRM into relatively distinct subsets. Tregs were enriched in CD103+ subsets. In line with this, Tregs’ functional markers, CD25 (IL-2Rα), PD-1, ST2 (IL-33R), ICOS (CD278), and CD39 were highly expressed within the CD103+CD69+ subset28 (Fig. 4f).

Fig. 4: Differential CD69 and CD103 expressions defined silicotic CD4+ TRM cells into effector or regulatory TRM cells.
figure 4

a Flow plot indicated CD4+ TRM cells in fibrotic lungs were divided into four subsets by CD69 and CD103 expressions. CD69+CD103+ (Red); CD69+CD103 (Blue); CD69CD103+ (Orange); CD69CD103 (Green). bd Flow histogram displayed transcriptional factor T-bet (b), ROR-γt (c), and GATA-3 (d) expressions among different subsets. Bar graph below shows the MFI among the indicated CD4+ TRM cell subsets. e Flow histogram compared ratios of FOXP3+ among different subpopulations. Bar graph below reveals the percentages of FOXP3+ among the indicated subsets. f FC heatmaps show the expression intensity of each marker. Representative flow histogram compared the intensity of CD25, PD-1, ST2, ICOS, and CD39 among four subsets in CD4+ TRM cells. g Typical flow plots showed CD69 and CD103 expressing patterns on Tregs (FOXP3+) or Teff cells (FOXP3) in CD4+ TRM cells. h Typical flow plots demonstrated CD69 and CD103 expressing patterns on T-bet+, ROR-γt+, or GATA-3+ subsets in CD4+ TRM cells. i The bar graph showed the ratios of CD103+ against CD103 subsets in CD69+ CD4+ TRM cells. Bar graphs are the combined results of at least three independent experiments. Individual mice are plotted on the graphs, n = 5–6 biologically independent animals. Values are reported as the mean ± SD. P value is determined by one-way ANOVA followed by Tukey’s test.

To corroborate the findings, we further divided the CD4+ TRM cells into Tregs and Teff cells, then checked their CD103 and CD69 expression patterns. Expectedly, TRM-Tregs (FOXP3+) expressed high CD103, while the TRM-Teff cells (FOXP3) expressed few CD103 (Fig. 4g), suggesting CD103 is a good indicator of TRM-Tregs. Furthermore, we discovered the Th1-type TRM (T-bet+) and Th17-type TRM (ROR-γt+) expressed less CD103, whereas Th2-type TRM (GATA-3+) expressed relatively high CD103 (Fig. 4h), implying the emergence of GATA-3+ Tregs29. Remarkably, the ratio of CD103+ to CD103 in CD69+CD4+ TRM cells was lifted with silicosis progression, emphasizing the immune imbalance within TRM subsets would be related to fibrogenesis (Fig. 4i). In synopsis, we proved the cell retention markers CD103 and CD69 can define CD4+ TRM cells into functional distinct lineages, which may provide potential therapeutic targets for alleviating silicosis.

Targeting CD103+ TRM-Tregs could not mitigate CS-induced pulmonary fibrosis

Next, we aim to explore whether targeting the CD103+ TRM subset exerts protective effects on silicosis progression. Taking the notion that CD103+ TRM-Tregs were preferentially recruited from the circulation, we employed FTY720 plus CD103 neutralization treatments to deplete the TRM subsets (Fig. 5a). Significantly, the treatment diminished CD103+ TRM subsets, resulting in a higher CD103/CD103+ ratio (Fig. 5b). Unexpectedly, we observed a semblable phenotype of the silicotic lungs, suggesting anti-CD103 treatment did not exert protective roles (Fig. 5c, d). Furthermore, we observed surged ratios of ROR-γt+ (Th17-type TRM) cells, while T-bet+ (Th1-type TRM) counterparts were not significantly affected (Fig. 5e, f). Significantly, depleting CD103+ TRM-Treg augmented Ki-67 levels of the CD103 TRM subsets (most TRM-Teff cells) (Fig. 5g), implying depleting CD103+ TRM cells unleashed TRM-Teff cell proliferation30. Collectively, these results demonstrated targeting CD103+ TRM cells could not mitigate CS-induced pulmonary fibrosis, which was related to the expansion of TRM-Teff cells (CD103 TRM).

Fig. 5: CD103+ TRM-Tregs exerted immuno-suppressive but not pro-fibrotic roles in the progression of silicosis.
figure 5

a Experiment design for CD103 neutralizing: silicotic mice treated with FTY720 (D28– D56 20 μg i.p.) together with CD103 neutralizing or isotype antibodies (100 μg/time i.v.) at indicated time points. b FC analysis of CD4+ TRM cells for CD69 and CD103 expressions. The bar graph compared the ratios of CD103 to CD103+ in the CD69+ CD4+ TRM cells. c and d H&E staining to the lung sections of silicotic mice treated with FTY720 together with CD103 neutralizing or isotype antibodies. Scale Bar = 500 μm (left) and 200 μm (right). e and f FC analysis of the percentages and numbers of T-bet+ or ROR-γt+ in pulmonary CD4+ TRM cells. g Flow histogram compared MFI of Ki-67 in the CD4+ TRM-Teff cells. Bar graph displayed the MFI. h The schematic diagram describes the process of CD103+ TRM-Tregs (CD4+ CD44+ FOXP3YFP+ CD103+) sorted in 8 weeks post-CS-treated FOXP3YFP mice by the FACS method. CS-pre-challenged mice were i.v. transferred with the sorted CD103+ Tregs or PBS. The recipient mice were analyzed 14 days after cell transfer. i H&E staining to the lung sections of distinct groups. Scale Bar = 600 μm (left) and 300 μm (right). The graph shows the inflammation score. j Masson’s trichrome staining to the lung sections of different groups. Scale Bar = 600 μm (left) and 300 μm (right). The bar graph showed the fibrosis-positive area. k Relative mRNA levels of Col1a1 and Fn in each group. l, m Relative mRNA levels of Il17a, Ifng, Tnfa, Il10 (l), and Tgfb1 (m) in each group. n FC analysis of the ratios and counts of IL-17A+ and IFN-γ+ in CD4+ T cells. For panels ag, n = 4–5 biologically independent animals. For panels in, littermate mice were used, n = 4 biologically independent animals. N number also indicated independent experimental replicates. Individual mice are plotted on the graphs. The black line indicates the data were analyzed in parallel. Values are reported as the mean ± SD. P value was determined by unpaired two-tailed Student’s t-test or paired two-tailed Student’s t-test.

CD103+ TRM-Tregs exerted immuno-suppressive but not pro-fibrotic roles in the progression of silicosis

To study the role of CD103+ TRM Tregs in silicosis, we depicted its immunophenotype compared to the CD103 counterparts by flow analysis. We observed higher IL-10 but not TGF-β (detected by LAP, essential for TGF-β cleaved) expressions in the CD103+ Tregs (Fig. S2a), demonstrating potent immunosuppressive property but not tissue-repair phenotype30. Moreover, there were more effector (PD-1+CD25+) and terminally differentiated (KLRG1+ICOS+) phenotypes in CD103+ Treg subsets (Fig. S2b, c)31,32. Additionally, we observed higher CD39 and CD69 levels in the CD103+ Tregs (Fig. S2d), suggesting high immune-suppressive and tissue-resident properties of the CD103+ subsets. These results indicated the CD103+ Tregs exerted high immune suppressive potentials.

To directly prove the role of CD103+ Tregs in the progression of silicosis, we did cell transfer experiments30. The CD103+ Tregs were sorted from the silicotic lung of FOXP3-YFP mice and transferred into another CS-pre-treated mouse (Fig. 5h). By histological analysis of H&E and Masson trichrome staining to the lung sections, we observed decreased inflammatory leukocyte infiltration (Fig. 5i), but unaffected collagen deposition (Fig. 5j). We further did qPCR analysis to the lung, and confirmed that Col1a1 and Fn transcripts were unaffected (Fig. 5k), indicating that CD103+Tregs did not exert profibrotic roles. While the pro-inflammatory mediators Il17a, Ifng, and Tnfa were descended, the immune suppressive cytokine Il10 was lifted in the lung tissue transferred with CD103+ Tregs (Fig. 5l). In contrast, the cell transfer did not affect Tgfb1 transcripts, further evidencing that CD103+ Tregs did not exert profibrotic roles (Fig. 5m). Furthermore, we detected proinflammatory cytokine productions of CD4+ Teff cells. It was demonstrated that appending CD103+ Tregs significantly reduced IL-17A and IFN-γ-producing Teff cells in the lung (Fig. 5n), but did not affect IL-13 production (Fig. S2e). Collectively, these results validated that CD103+ Tregs functioned as an immunosuppressive regulator restraining Teff cells rather than profibrotic mediators.

Neutralizing IL-7 in lung retarded silicosis progression through disrupting the pathogenic TRM-Teff cell maintenance

Now that the CS-induced pathogenic CD4+ TRM cells expanded in situ, we next explored interventions targeting their maintenance in the lung, in which IL-7 was reported to be essential33. We first examined the IL-7R (CD127) levels on the CD4+ TRM cells in silicotic lung. Expectedly, CD4+ TRM cells expressed higher levels of IL-7R than the naive T cells (CD44). Strikingly, within CD4+ TRM cells, TRM-Teff cells, but not TRM-Tregs, expressed a higher level of IL-7R, indicating a high demand for IL-7 of those Teff cells (Fig. 6a). Accordingly, the surged level of Il7 expression in the silicotic lung was confirmed (Fig. 6b). We then aimed to treat the silicotic mice with IL-7-neutralizing antibody. Since IL-7 was critical in lymphocyte maintenance, we chose to treat the mice through intratracheal (i.t.) instillation to avoid affecting the immune response in other organs (Fig. 6c). We sorted CD4+ T cells from the lungs and spleens, then did qPCR analysis. Significantly, in the treated pulmonary CD4+ T cells, we observed alleviated Th1 and Th17-related transcripts (Tbx21, Ifng, Il2, Rorc, and Il17a), as well as decreased mRNA levels associated with cell activation (Icos, Ctla4, Klrg1, and Pdcd1). Additionally, the cell retention markers (Cxcr6 and Itgae) were decreased, implying a reduction of TRM cells (Fig. 6d). Strikingly, the anti-IL-7 treatment augmented Il10 transcripts, while the markers related to Tregs, Foxp3, and Areg were unaltered (Fig. 6d). However, these effects were diminished in the splenic CD4+ T cells (Fig. 6e), suggesting that pulmonary local IL-7 neutralization did not affect immune response in whole bodies.

Fig. 6: Neutralizing IL-7 in the lung retarded silicosis progression through disrupting the maintenance of pathogenic TRM-Teff cells.
figure 6

a Flow histogram of IL-7R (CD127) on naive or TRM-Tregs and Teffs. The graph compared the MFI. b Relative mRNA levels of Il7a in mice lung tissue of saline- or CS-treated mice. c Bioluminescence of silicotic mice with isotype-control or IL-7 neutralizing (30 µg i.t. twice a week from D28). d RNA analysis of sorted pulmonary CD4+ T cells in distinct groups. The heatmap contained Th1, Th17, and Treg-related genes, activation, and retention markers. *P < 0.05; **P < 0.01. e RNA analysis of sorted splenic CD4+ T cells in distinct groups. The heatmap contained Th1, Th17, and Treg-related genes. f FC analysis of CD69 and CD103 expression in CD4+ TRM cells. The ratios of CD103 to CD103+ in the CD69+ CD4+ TRM cells were shown. g FC analysis of CD4+ TRM cells for T-bet and ROR-γt expression. The graph displayed the percentages of T-bet+ or ROR-γt+ TRM cells. h Flow histogram indicated the Ki-67 levels in the TRM-Teffs. The graph compared the MFI. i Relative mRNA levels of Ifng and Il17a in mice lung tissue of each group. j H&E staining to the lung sections of different groups. Scale Bar = 500 μm (upper) and 200 μm (lower). The inflammation scores were measured. k Masson’s trichrome staining to the lung sections of different groups. Scale Bar = 500 μm (upper) and 200 μm (lower). The fibrosis-positive areas were quantified. l Relative mRNA levels of Col1a1 and Fn in mice lung tissue of each group. For panel a, n = 5 biologically independent animals. P value was determined using one-way ANOVA followed by Tukey’s test. For panels bl, n = 5–6 biologically independent animals. Individual mice are plotted on the graphs. N number also indicated independent experimental replicates. P value was determined by unpaired Student’s t-test. Values are reported as the mean ± SD.

To further detect the effects on CD4+ TRM subsets, we did flow analysis and observed a decreased ratio of CD69+CD103 to CD69+CD103+ (Fig. 6f), indicating that the pathogenic CD69+CD103 subsets were restrained. By scrutinizing the composition of subsets within CD4+ TRM cells, we proved pulmonary IL-7 neutralization blunted T-bet+ Th1-type TRM, but not the ROR-γt+ Th17-type TRM cells (Fig. 6g). In contrast, the unaltered Treg ratio excluded its influences on TRM-Tregs, further confirmed by Ki-67 and CD25 levels (Fig. S3a, b). Notably, pulmonary IL-7 neutralization restrained TRM-Teff cell proliferation, manifested by decreased Ki-67 levels (Fig. 6h), confirming that IL-7 promoted the expansion of TRM-Teff cells34. We also did a PCR analysis of the lung tissues. The Th1-related cytokine (Ifng), as well as Th17-related cytokine (Il17a) transcripts, were decreased in the lung (Fig. 6i), which was in line with previous results. We performed lung histological analysis and validated that IL-7 neutralization reduced immune infiltration, cellular nodule formation (Fig. 6j), and attenuated collagen deposition (Fig. 6k). Further, the declined Col1a1 and Fibronectin transcripts in the cytokine-neutralized lung validated the alleviated fibrosis phenotype (Fig. 6l). Collectively, these data demonstrated that pulmonary local IL-7 intervention restrained CS-induced pulmonary fibrosis by disrupting the TRM-Teff maintenance and function, highlighting the important role of TRM-Teff in mediating silicosis.

Discussion

Here we showed that a substantial accumulation of pulmonary CD4+ TRM cells mediated the pathogenesis of silicosis, which exerted immunological memory and antigen-specific response to the invaded CS particles. The CD4+ TRM cells, especially CD69+CD103+ TRM-Tregs, rely on continual replenishment from circulating lymphocytes. The defined population by CD103 and CD69 displayed distinct functional phenotypes. Targeting the immunosuppressive CD103+ TRM cells did not exert protective roles to silicosis. Whereas neutralizing IL-7 in the lung disrupted the maintenance of TRM-Teff cells during silicosis progression and exerted protective effects.

Since the invaded CS particles could not be cleared by the pulmonary immune system, it would lead to repeated and cycled antigen stimulation, in which our one-time intratracheal CS instillation gave rise to CD4+ TRM cells. Rag1–/– mice transfer experiment and repeated CS exposure jointly revealed that new CS exposure induced rapid TRM cell expansion and potent reaction, demonstrating their immunologic memory and antigen-specific characterization. Notably, the CS particles could induce T-cell proliferation by directly activating cells through T-cell antigen receptor (TCR) complexes but not rely on antigen-presenting cells35. Additionally, the rearrangement of TCR was confirmed by TCR repertoire sequencing under silica particle exposure36. Further study about the structural or molecular mechanism of how CS particles activate T cells, for instance, exploring the CS binding site within the TCR footprint or the presenting MHC molecule and the peptide bound by the MHC, would provide insights into the disease mechanisms and targets for antigen-specific therapies.

By circulatory lymphocyte blockade and parabiosis experiment, we proved the circulating T cells contributed to expanded TRM cells, which confirmed that lymphocytes migrating from lymphoid organs to tissues are the origin of TRM cell formation12. Analog to our results, the circulatory T cells contribute to CD4+ TRM cells exacerbating asthma37. Significantly, we noticed circulating lymphocyte recruitment blockage by FTY720 decreased the percentage of Tregs in CD4+ TRM, implying their high demand for replenishment from circulating lymphocytes. Ultimately, our parabiosis experiment directly proved CD69+CD103+ TRM-Tregs depended more on circulatory T cell replenishment. In supporting the notion that we previously reported the TRM-Tregs in the silicotic lung expressed high Nrp-1 (markers of thymus-derived Tregs)23,38, further proving the TRM-Tregs originally came from the thymus and recruited to peripheral tissue, but not the conversion of Teff cells to Tregs39. All the evidence supported the suggestion that circulating thymus-derived Tregs are required to replenish CD69+CD103+ TRM-Tregs to ensure a tissue-local immunosuppressive environment.

Remarkably, even though FTY720 treatment could attenuate silicosis following our results, it should be cautious. A common adverse effect would be an increased incidence of respiratory infections because of systemic immune cell deficiency40. Hence, we further explore the subsets of CD4+ TRM cells, which could provide more available interventions. The CS-stimulated pulmonary CD4+ TRM cells expressed CD69, regardless of Teff cells or Tregs, implying a crucial role of this molecule in silicosis. However, research reported genetically deleting CD69 in mice did not affect CD4+ TRM cells41, indicating that CD69 per se is not necessarily sufficient for influencing TRM cell maintenance or function. Analogous to the immuno-suppressive CD69+CD103+ Tregs in silicosis, CD69+ Tregs protect from inflammatory damage after myocardial infarction42. Surprisingly, the prominent expression of CD103 was also confirmed on cytotoxic TRM cell subsets, like the CD103 subsets in silicosis43,44. We speculated that tissue microenvironment and disease model contributed to the distinction of TRM cell phenotypes45. Thus, a comprehensive understanding of the effects of CD69 and CD103 on the function and immuno-balance within CD4+ TRM cells still needs dissection.

Despite numerous studies exploring the role of Tregs in fibrotic disease, this topic is still debatable46,47, related to Tregs’ heterogeneity in peripheral tissues48,49. In pursuing the role of circulatory-derived CD103+ TRM-Tregs in silicosis, we applied FTY720 together with CD103 neutralizing antibody to the silicotic mice, expecting to get an alleviated effect. Surprisingly, this treatment did not mitigate disease phenotype. While the restrained effector T proliferation was unleashed by CD103 neutralization, consistent with the notion that CD103+ TRM-Tregs restrained pathogenic CD4+ Teff cells in the lung50. By direct cell transfer experiment, we demonstrated CD103+ TRM-Tregs limited proinflammatory cytokine productions and exerted inhibitory effects on chronic inflammation, but could not alleviate fibrosis. Because the replenishment of TRM-Tregs was highlighted by our parabiosis experiment, a shortcoming of our study was that one-time cell transfer might not maintain long-lasting effects of CD103+ TRM-Tregs against fibrogenesis. Furthermore, research about markers distinguishing pro-fibrotic or immunosuppressive Tregs would provide insights into treating silicosis.

Specifically, CD69+CD103 TRM possessed a robust proinflammatory response to invaded particles manifested as pathogenic Th1 and Th17-type TRM-Teff cells51, which was in line with the observation of vaccine or transplantation‐induced Th1-like CD69+CD103 TRM subsets26,52. The invaded CS particles damaged lung tissue and triggered inflammatory responses, while fibrogenesis was pathological excessive tissue repairment. We found that CD69+CD103 CD4+ TRM cells expanded in response to repeated CS stimulation, secreting pro-inflammatory mediators IFN-γ and IL-17A53,54. Since the particles could not be cleared from lung tissue, repeated CS stimulation would lead to proinflammatory TRM cell expansion and enlarge lung local damage, eventually promoting the progression of silicosis55. Thus, modulating immuno-balance within CD4+ TRM cells that restrained the number and function of CD69+CD103 TRM subsets may provide therapeutic effects in treating silicosis.

IL-7 are crucial cytokines produced by pulmonary fibroblasts and epithelial cells in fibrosis, regulating lymphocyte development and maintaining TRM cell viability56,57. Herein, we illuminated TRM-Teff cells expressed higher IL-7R, indicating the critical role of IL-7 in cell maintenance and function. Higher cytokine level of IL-7 in the serum was observed in silicosis patients58, whose role in silicosis was elusive. Although the suppressive role of IL-7 on TGF-β signaling has been validated59, the maintenance of pathogenic TRM subsets via IL-7 could also aggravate tissue injury and subsequent pulmonary fibrosis, by producing other pro-inflammatory cytokines. We demonstrated that IL-7 neutralization in the lung was effective in alleviating CS-induced pulmonary fibrosis by decreasing pathogenic CD4+ TRM-Teff cells but not TRM-Tregs expressing low IL-7R. Neutralizing IL-7 locally in the lung disrupted the maintenance of TRM-Teff cells, attenuated CS-induced pulmonary chronic inflammation, mitigated local inflammatory damage to lung tissue, and thus inhibited fibrosis progression (excessive tissue repair). In particular, the Th9 cells might reside in TRM-Teff subsets with higher IL-7R expression, which was a major player in inducing fibrosis60. Since Th9 cells were involved in IL-7R high TRM-Teff subsets, the anti-IL-7 treatment could successfully attenuate the function of Th9 cells, overlapping the effect of anti-IL-9 treatment. Reversely, the successful postponed pulmonary fibrosis under anti-IL-7 blocking might also contain the decreased maintenance of Th9 cells. We did not utilize neutralizing antibodies by intravenous treatment since it may affect immune responses in other organs. Expectedly, lung local IL-7 neutralization did not affect the immune response in the spleen. Besides, bronchoalveolar lavage appending IL-7 neutralizing treatment may provide postponing effects to silicosis in the advanced stage.

In conclusion, our results provided novel functions of CD4+ TRM cells in silicosis and new insights into CS particle’s toxicological effects in the lung, all of which will be essential for the development of new therapeutic strategies for postponing the intractable inflammation-associated fibrotic diseases.

Methods

Mice

C57BL/6JGpt mice (Strain No. N000013), CD45.1 congenic mice (C57BL/6JGpt-Ptprcem1Cin(p.K302E)/Gpt, Strain No. T054816), Rag1–/– mice (C57BL/6JGpt-Rag1em1Cd3259/Gpt, Strain No. T004753) involved in this study were purchased from GemPharmatech (Nanjing, China). FOXP3YFP mice (B6.129(Cg)-Foxp3tm4(YFP/icre) Ayr/J, Strain #016959) were purchased from Jackson laboratory. All mice were bred and maintained under specific pathogen-free conditions in the animal care facility at China Medical University. Sex- and age-matched mice were used. All mice were used at the age of 7–8 weeks. Littermate mice were selected if possible. The experimental protocols were approved by the Institutional Animal Care and Use Committees (IACUC) of China Medical University and all animal experiments were performed in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals.

Crystalline silica particles

Crystalline silica (CS) particles were purchased from the U.S. Silica Company (Frederick, MD, USA). The characteristics of crystalline silica particles are described in detail23. Briefly, the size distribution of CS particles is as follows: 97% <5 μm in diameter, 80% <3 μm in diameter, median diameter of 1.4 μm. The particles were suspended in sterile saline after drying. The suspension was autoclaved and sonicated for 10 min before use.

The murine model of silicosis establishment

The mouse model of silicosis was established according to our previously published method23. In brief, the mouse was treated with 3.0 mg CS in 50 μL saline solution by intratracheal (i.t.) instillation after anesthetization with pentobarbital sodium (30 mg/kg, i.p. injection, Sigma). An equal amount of sterile saline was applied to the control groups. 56 days post-CS instillation was regarded as the stage of fibrogenesis. Besides, the mouse in the CS repeated exposure group was treated with 1.0 mg CS in 50 μL saline solution per week by i.t. instillation for successive three weeks.

Single-cell suspension preparation

Isolation of lymphocytes from mouse lungs and spleen was previously described23. In short, the lung was minced and digested with a digestion solution containing type I collagenase (2 mg/mL), DNase I (100 U/mL), and complete media (DMEM plus 4% BSA). The suspension was incubated on a rocker at 37 °C. Collagen-digested lungs were dispersed. Red blood cells were lysed using ACK lysis buffer. Leukocytes were enriched from the cell digestion using Percoll gradients (80%/40%) (Cytiva). Spleen cells were prepared by pressing the tissues through 70 μm cell strainers using the end of a sterile plunger of a 5 mL syringe. Single-cell suspension was prepared for subsequent flow cytometry staining or cell sorting.

Flow cytometry analysis

Flow cytometry (FC) was performed according to the guidance61. Live/Dead cell viability dye Aqua (Invitrogen) was used to exclude the influence of dead cells. Fluorescently labeled antibodies to cell surface antigens were applied and incubated at 4 °C for 30 min. For the intracellular transcriptional factor staining, cells were first stained for cell-surface markers and fixed with 4% paraformaldehyde, permeabilized with the FOXP3 transcription factor staining buffer set (eBioscience), and then stained with indicated antibodies to transcriptional factors at 4 °C overnight. For the cytokine staining, the leukocyte stimulation cocktail containing PMA, ionomycin, brefeldin A, and monensin (Invitrogen) was used to stimulate cells for 4 h before staining. The antibodies detecting transcriptional factors or cytokines were incubated at 4 °C overnight. FC analysis was performed on BD FACS Celesta (BD Bioscience, San Jose, CA, USA). FACS data were analyzed with Flowjo 10.6 software. The gating strategy is shown in Fig. S1a. All involved anti-mouse antibodies are listed in Table S1.

Intravascular immune cell labeling

To discriminate tissue-resident cells from circulating cells, we carried out a well-established intravenous (i.v.) staining approach, illuminated in Fig. 1a23. Mice were i.v. injected with 1.5 μg of CD45-APC-Cy7 antibody (30-F11, Biolegend) diluted in 150 μL sterile saline, 4 min before sampling.

Magnetic-activated cell sorting and adoptive transfer

CD4+ αβT cells were purified from the tissues of 8 weeks post CS-treated lung, 8 weeks post CS-treated spleen, and saline-treated spleen by the magnetic-activated cell sorting (MACS) method with a CD4+ T cells Isolation Kit (Miltenyi Biotec, Cat#130-104-454). The purity of the isolated cells was >95%. The MACS-purified CD4+ T cells or PBS were i.v. transferred into littermate Rag1–/– mice (3 × 105 cells per mouse). After that, CS was instilled. One week after the CS treatment, the transferred mice were analyzed. The donor and recipient mice were sex matched.

Flow cytometric cell sorting and adoptive transfer

FOXP3YFP+ mice were treated with CS particles. The activated CD103+ Tregs (CD4+ CD44+ FOXP3YFP+ CD103+) were sorted from the lung of 8 weeks post-CS-treated FOXP3YFP+ mice by flow cytometric cell sorting (FACS Aria II instrument, BD Biosciences). The gating strategy is shown in Fig. 5h. The sorted Tregs (1 × 104 cells each mouse) or PBS were i.v. transferred into littermate C57BL/6J mice treated with CS 4 weeks before transfer. Two weeks later, the reconstituted mice were analyzed. The donor and recipient mice were sex matched.

Parabiosis model

The experiment was adopted as the published protocols26,27. Briefly, sex- and age-matched congenic mice (CD45.1/2 and CD45.1/1) were co-caged and prepared. Their dorsal and ventral skin were approximated with sutures after parabiosis surgery. The conjoined mice recovered for 14 days to establish new circulation. Then, CD45.1/1 mice were CS administered, and sacrificed 7 days later for flow analysis.

RNA extraction and quantitative PCR (qPCR) analysis

RNA was extracted from lung tissue by RNA isolater Total RNA Extraction Reagent (R401, Vazyme Biotech), then reversely transcribed into cDNA by HisScript IIIs RT SuperMix (R323, Vazyme Biotech). The ChamQ Universal SYBR qPCR Master Mix (Q711, Vazyme Biotech) was used for the amplification of RNA samples from each group and gene expression was analyzed via real-time PCR assay (7500 software, Applied-biosystem). GAPDH was used as the internal control for determining 2−ΔΔCT values. All primers were synthesized by Sangon Biotech, and the sequences are listed in Table S2.

FTY720 treatment

FTY720, an agonist of sphingosine-1-phosphate receptor 1 (S1PR1), was used to block lymphoid cell migration. FTY720 (20 μg in the vehicle every time) or vehicle was i.p. administered daily (Fig. 3a). Littermate mice were used in each independent experiment. The mice were divided into three groups: (A) CS-treated group, in which animals received CS instillation and vehicle; (B) Full-time FTY720 treatment (0–8 weeks), in which the mice received CS instillation combined with FTY720 i.p. every day from 0 to 8 weeks; (C) Half-time FTY720 treatment (4–8 weeks), in which the mice received CS instillation combined with FTY720 i.p. every day from 4 to 8 weeks.

Neutralizing antibody treatments

Anti-CD103 treatment: Anti-mouse CD103 (M290, BE0026) and isotype-matched antibodies were purchased from Bio X cell. C57BL/6J mice were divided into two groups (Fig. 5a): Isotype group, in which mice received CS instillation and FTY720 treatment (D28 to D56 every day) together with isotype antibody (100 µg i.v. D36– D56 every other day). Anti-mouse CD103 group, in which the mice received CS instillation and FTY720 treatment (D28 to D56 every day) and anti-mouse CD103 antibody (100 µg i.v. D36 to D56 every other day). Anti-IL-7 treatments: Anti-mouse IL-7 antibody (M25, BE0048) and isotype-matched antibody were purchased from Bio X cell. Briefly, C57BL/6J mice were divided into an anti-IL-7 group and an Isotype group, in which mice received CS instillation and isotype or anti-mouse IL-7 antibody (30 µg, i.t., twice a week from D28 to D56).

Lung histological analysis

The lung tissues of mice were fixed with 4% paraformaldehyde (PFA) and embedded in paraffin. Slices (4 μm thick) were cut, mounted on slides, and stained. Hematoxylin & Eosin (H&E) were conducted to assess the degree of inflammation and pathological changes. Collagen fiber content was quantified using Masson’s trichrome staining. The staining was performed according to the manufacturer’s protocol. Stained lung sections were photographed in a microscope (Leica M205 FA, Wetzlar, Germany). The inflammation score was done by a semi-quantitative analysis based on a previously published method23. Briefly, lung inflammation was graded into four stages and scored as follows: normal lung, 1 point; light inflammation, 20% of lung area, 2 points; medium inflammation, 20–50% of lung area, 3 points; severe inflammation, 50% of lung area, severe structure distortion, 4 points. The fibrotic area is presented as a percentage, measured by Image J.

Statistics and reproducibility

It was assumed that sampling was from a normally distributed population. Statistical analysis was performed using GraphPad Prism 9.0.1 software. Unpaired/paired two-sided Student’s t-test (2 groups) or one-way analysis of variance (ANOVA) followed by Tukey’s test (more than 2 groups) was used to evaluate the statistical differences between groups. All data were presented as the Mean ± SD for bar graphs unless otherwise stated, with sample size (n) specified in figure legends. Absolute P values are indicated in each figure. For all analyses, a P value < 0.05 was considered to indicate significance. Experimental results were reliably replicated. For each experiment, the number of replications was noted in figure legends.

Study approval

The experimental protocols were approved by the Institutional Animal Care and Use Committees (IACUC) of China Medical University, and we have complied with all relevant ethical regulations for animal use, in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.