Selective sexual dimorphism in musculoskeletal-cardiopulmonary pathologies and mortality in the TNF-transgenic mouse model of rheumatoid arthritis

Abstract

Objective:

To examine and quantify the sexual dimorphism in musculoskeletal and cardiopulmonary pathologies in the TNF-Tg (Tg3647) mouse model of inflammatory-erosive arthritis.

Methods:

Kaplan-Meier estimates of male and female Tg3647 and WT littermates were determined. Longitudinal and cross-sectional outcomes of musculoskeletal and cardiopulmonary pathologies were assessed via ultrasound, μCT, grip strength, histology, serology, flow cytometry and skeletal muscle physiology.

Results:

Female Tg3647 mice have significantly shorter lifespans with: 1) gross deficits in body mass and muscle weight, 2) early onset inflammatory arthritis with similar end-stage severity, 3) early onset and increased severity of inflammatory interstitial lung disease (ILD). Histologically, the ILD was characterized by inflammatory cell accumulation and pulmonary arteriole thickening, which was concomitant with right ventricular hypertrophy that was also more severe in female versus male Tg3647 mice. No sexual dimorphisms in TNF-induced deficient: grip strength, axial skeletal growth, or bone loss were found. Globally, the extent of the pathologies studied in Tg3647 versus WT females was greater than that observed in Tg3647 versus WT males.

Conclusions:

We find that TNF selectively drives early onset arthritis and cardiopulmonary pathologies in females. These results in the Tg3647 mouse identify it as a suitable model to better understand the mechanisms underlying sexual dimorphism and cardiopulmonary disease in the setting of inflammatory arthritis and other connective tissue diseases.

Introduction

Autoimmune diseases, such as Rheumatoid Arthritis (RA), Systemic Lupus Erythematosus, Grave’s disease, Sjogren Syndrome and Scleroderma, exhibit remarkable sexual dimorphism with female: male incidence ratios of at least 2:1 and some of these diseases are ten times more frequent in females [1]. RA is one of the most prevalent of these autoimmune diseases, affecting 0.6–0.8% of the adult population in the US, second only to thyroid disease [2, 3]. Conceptually, three main factors may account for the predominance of these disorders in females: steroid hormones, chromosomal differences based on the number of X chromosomes, and environmental or societal differences [1]. The analysis of environmental or societal differences in pre-clinical models presents major obstacles, and a significant body of work has focused on the interaction between autoimmunity and sex hormones and chromosome linked traits [1, 4–6]. Of particular relevance, however, is the absence of current models to study the role of sexual dimorphism in the development of inflammatory arthritis and concomitant extra-articular features [7].

The tumor necrosis factor transgenic mouse (TNF-Tg) is a model of multiple diverse inflammatory phenotypes [8–10]. Specifically, the Tg197 and Tg3647 strains possess the full-length human TNF gene with a modified 3’UTR exchanging for the β-globin 3’UTR and develop spontaneous inflammatory erosive arthritis at one and three months, respectively [8, 10]. This difference in disease onset is thought to be due to the number of transgenes inserted, as the Tg197 carries ~five copies and the Tg3647 carries one copy [8, 10]. However, differences in the timing of disease onset and severity between male and female mice have not been fully explored. Additionally, other murine models of spontaneous arthritis, such as the SKG, K/BxN and other TNF induced arthritis models (i.e. models driven by different TNF transgenes), have not been described to display sexual dimorphism [7, 11].

Herein, we expand upon anecdotal evidence within our lab that female TNF-Tg (Tg3647 line) mice have shorter lifespans and develop arthritis earlier than males. These studies were motivated by recent epidemiologic evidence describing increased mortality, earlier arthritis onset and increased arthritic severity in female patients with RA [12–14]. We investigated the mortality incidence of Tg3647 mice with supporting necropsy analysis of female mice, which demonstrated that their dramatically shortened lifespan may be related to cardiopulmonary disease compared to male littermates. We also performed longitudinal and cross-sectional analyses of the cardiopulmonary and musculoskeletal systems to investigate sexual dimorphism of both rheumatologic disease progression and pathophysiology related to the increased mortality. We found that female Tg3647 mice manifest arthritic symptoms at earlier time points than males along with development of severe cardiopulmonary disease that explains the high mortality incidence and shorter lifespan in the female mice.

Materials and Methods

The present work was conducted with prior approval of the University of Rochester Medical Center (URMC) University Committee for Animal Resources (UCAR). The 3647 Tumor Necrosis Factor transgenic (Tg3647) mouse line was originally obtained from Dr. George Kollias [8, 10] and has been maintained across multiple generations. All studies were performed with littermate wild-type controls (WT). An initial cohort of WT male (n=28), WT female (n=27), Tg3647 male (n=30) and Tg3647 female (n=34) mice were observed for 400 days under standard microisolator conditions with autoclaved food (Rodent Diet 5010, LabDiet St. Louis, MO) and water available ad libitum. Date of birth and date of death were recorded for Kaplan-Meier estimates. Subsequently, four Tg3647 female mice at six months of age were submitted for internal organ necropsy analysis. From this analysis a prospective study was designed and performed to assess longitudinal outcomes from 2 to 5.5 months of age, as well as cross sectional timepoints at 3, 4 and 5.5 months (1 month = 28 days). Therefore, three cohorts of four groups (n=6 each) composed of WT male, WT female, Tg3647 male and Tg3647 female mice were created with each cohort being sacrificed at either 3, 4 or 5.5 months of age (Supplemental Figure 1). Confirmatory studies with additional mice were performed to assess complementary outcomes as needed. Detailed methods for histology, ultrasound, micro-Computed Tomography (μCT), grip strength, muscle force generation, tibialis anterior (TA) dissection, spine analysis, flow cytometric analysis, cytokine levels and statistics are presented in the Supplemental Document.

Results

Earlier Mortality in Female Tg3647 Mice with Concomitant Pathology within the Cardiopulmonary and Musculoskeletal Systems

Kaplan-Meier estimates of female and male Tg3647 and WT mice revealed a significant decrease in the median lifespan of female Tg3647 mice compared to their male counterparts (Figure 1A, 166 vs. 229 days, p<0.001). In order to understand the etiology of the accelerated mortality in transgenic females, an internal organ necropsy study was performed on four female Tg3647 mice that were euthanized at six months of age. This investigation revealed significant cardiopulmonary pathology in all mice, including marked numbers of plasma cells, lymphocytes, and macrophages surrounding pulmonary vessels and airways, as well as thickened right ventricles (Supplemental Document). These unexpected findings triggered a comprehensive assessment of the temporal progression of the cardiopulmonary and musculoskeletal pathologies in the male and female Tg3647 mice and their WT littermates. The purpose of these studies was to formally define the female-male temporal pathologic divergence in this model of RA. Of note, no remarkable pathologies were identified in liver, kidney, large intestine, small intestine, brain and salivary glands in any mice (Supplemental Document), but pathologic analysis did reveal the known musculoskeletal and splenic abnormalities well established in the Tg3647 mice [8–10, 15].

Figure 1. Increased morbidity and accelerated mortality in Tg3647 female mice compare to male Tg3647 mice.

Female Tg3647 have a significantly shorter lifespan, with a median survival expectancy of only 166 days compared to their male counterparts of 229 days (A, p<0.001). Female Tg3647 mice have lower total body weight than sex matched littermate and male Tg3647 counterparts and lose weight from 3 to 5.5 months of age (B). Male and female Tg3647 both had significantly decreased grip strength compared to their sex matched WT counterparts at all time points (C). Absolute EDL muscle force showed severe deficits in female Tg3647 (D), however this was significantly contributed to by the lower muscle mass in female Tg3647 mice (E). When normalized to cross sectional area of the muscle, the specific force lacks any difference between the groups (F). Statistical results: † Tg3647 female different from TNF male p<0.05, § Tg3647 female different from all groups p<0.05, ¶ Tg3647 male different from WT male p<0.05, ‡ Tg3647 female different from WT female p<0.05, ***p<0.001.

Specific Musculoskeletal Sexual Dimorphism in Tg3647 Mice

Total body weights of both male and female Tg3647 mice were significantly decreased compared to their sex matched WT littermates at 4, 5 and 5.5 months (Figure 1B). Importantly, female Tg3647 mouse weight decreased from 3 to 5.5 months of age, while there was no difference in male Tg3647 or WT mouse weight during this period. No differences in bone growth were observed until 4 months of age when differences between the L5 height of Tg3647 and WT mice became apparent (Supplemental Figures 2A–E). Evidence of generalized osteopenia in Tg3647 vs. WT mice was similar to that previously reported [16]; no differences between male and female Tg3647 mice in L5 trabecular BV/TV and cortical bone volume were found (Supplemental Figure 2F–O).

Grip strength was significantly decreased in both male and female Tg3647 mice at all time points compared to sex-matched littermate controls (Figure 1C). At 2 and 3 months of age, female Tg3647 mice had significantly reduced grip strength compare to age-matched male Tg3647. However, post hoc analysis revealed no interaction of sex and genotype within these time points, suggesting female mice have less grip strength independent of genotype (Figure 1C). In terms of skeletal muscle, absolute extensor digitorum longus force showed severe deficits at 5.5 months, however, this was primarily due to the lack of EDL muscle mass (Figure 1D and E). When normalized accounting for cross sectional area was performed, there were no differences in specific EDL muscle force (Figure 1F).

Arthritic changes in the knee occurred earlier in female Tg3647 mice compared with males. The females demonstrated significantly increased total synovial area, cells within the synovium, total histology score and inflammatory infiltrates at three months of age (Figure 2 A–F, Supplemental Figure 3 A–J). In contrast, the severity of ankle arthritis was similar between the sexes at three months of age (Supplemental Figure 4 A–D). Wild type knee and ankle histology remained normal over the study period (data not shown). Analysis of the popliteal lymph node (PLN) volume, a biomarker of knee arthritic changes [15, 17, 18], also showed significant alterations at earlier time points in female Tg3647 compared to male Tg3647 mice. While both male and female Tg3647 mice experienced a progressive increase in their PLN volumes at the same rate, vascularity of female PLNs was greater than males at three months of age as determined by normalized power Doppler volume (NPDV). Female total LN volume began to decrease earlier than males at 5.5 months of age (Figure 2 G and H). The earlier changes in female PLN dynamics was clearly demonstrated when the classification scheme was applied to categorize the LNs into expanding and collapsed phases (Supplemental Table 1). Significantly more female Tg3647 expanding PLNs (early arthritis) were identified compared to male Tg3647 PLNs at three and four months of age. Further, eight female PLNs collapsed (advanced arthritis) by 5.5 months, while only three male PLNs collapsed at 5.5 months. Lastly, the trends of each PLN volume individually showed that female PLNs reached peak volume and began to collapse much earlier than the majority of male PLNs (Figure 2 I and J). Overall, these data indicated that Tg3647 female mice developed earlier knee arthritis and joint draining lymph node pathology and significant musculoskeletal morbidity compared to their male counterparts.

Figure 2. Female Tg3647 mice have earlier onset of knee arthritis and joint draining lymph node dysfunction.

At 3 months of age male Tg3647 mice (A) have significantly less arthritic change in their knees than their female counterparts (B, Arrows). Histomorphometry at each cross-sectional time point shows a significant increase in total synovial area (C) and synovial cellular area (D) in female Tg3647 at 3 and 4 months of age compared to male Tg3647. Further, the total histology score (E) from blinded scoring on inflammatory synovial infiltrate, pannus invasion and TRAP+ area demonstrate a considerable increase in pathology at both 3 and 4 months with inflammatory infiltrate playing a major role (F). Popliteal lymph node size (G) and blood flow (normalized power Doppler volume, H), biomarkers of knee arthritic progression, do not show appreciable changes in male and female WT over time. However, both Tg3647 male and female mice demonstrate an increase in their volume and NPDV, with the female Tg3647 NPDV peaking earlier than male counterparts at 3 and 4 months. This is further shown by plots of each individual LN volume over time for female Tg3647 (I) and male Tg3647 (J). Statistics: ***p<0.001; *p<0.05; NS = Not Significant.

Accelerated and More Severe Cardiopulmonary Disease in Female Tg3647 Mice.

Although prior reports have documented the interstitial lung disease (ILD) [19, 20] in Tg3647 mice, sexual dimorphism of pulmonary related pathology has not been studied. Thus, we performed a comprehensive assessment of lung and heart in longitudinal and cross-section cohorts of Tg3647 vs. WT, male and female mice described above. Representative images of lung sections demonstrated dramatic differences at 4 months between age-matched male and female mice including increased perivascular and peribronchiolar inflammatory infiltrates as well as thickened alveolar septums and interstitial infiltrates in females (Figure 3 A, B, D, E; Supplemental Figure 5A–D). When histomorphometry was performed to calculate percentage of cell area within the lung tissue sections, we observed a striking increase in female Tg3647 mice at 4 and 5.5 months of age compared to male transgenics (Figure 3C). Blinded scoring of each section for specific pathologic features indicated a similar relationship in which female Tg3647 mice had significantly increased total histology scores compared to male Tg3647 mice at 4 and 5.5 months of age (Figure 3F). When investigating the sub-scores, peribronchiolar infiltrates (4mo: 0.7 ± 0.3 vs 1.4 ± 0.5; 5.5mo: 1.0 ± 0.1 vs 2.1 ± 0.6), perivascular infiltrates (4mo:1.2 ± 0.4 vs 2.0 ± 0.6; 5.5mo: 1.2 ± 0.4 vs 2.3 ± 0.5) and arteriole thickness (4mo: 0.9 ± 0.3 vs 2.1 ± 0.6; 5.5mo: 1.1 ± 0.5 vs 2.3 ± 0.4) contributed most to the differences between male and female Tg3647 at 4 and 5.5 months of age, respectively (Supplemental Table 2). In vivo lung μCT measurement of aerated volume, tissue volume and total volume confirmed the massive cellular burden of female Tg3647 with a significant decrease in air volume at 4 and 5.5 months of age, and a significant increase in tissue volume at 4 months of age (Figure 3 H–J). Interestingly, interstitial fibrosis was not significant when scored pathologically (data not shown) nor when analyzed directly with histomorphometry (Supplemental Figure 5 E and I). However, the increased arteriole thickness score and intense Masson staining around the small arteries in female transgenic mice warranted further investigation.

Figure 3. Increased interstitial and vascular pulmonary pathology in female Tg3647 mice.

Representative H and E lung micrographs of four-month-old male (A) and female (B) Tg3647 depict exacerbated pulmonary pathology in the female mice. Histomorphometry reveals significantly more nuclei area at 4 and 5.5 months of age in Tg3647 female lungs (C). High magnification images of male (D) and female (E) Tg3647 show increased peribronchiolar (Green Arrow), perivascular (Black Arrows) and interstitial infiltrates (Blue Arrows), as well as dramatic arteriole thickening (Red Arrow). Blinded scoring of these features demonstrates a significant increase in total histology score at 4 and 5.5-months of age Tg3647 female lungs compared to males and WT littermates (F). μCT shows a significant reduction in air volume in lung (4 and 5.5 months of age) and significantly more tissue and total volume (4 months of age) in female Tg3647 mice compared to male Tg3647 mice (G-H). Statistical results: † Tg3647 female different from TNF Male p<0.05, § Tg3647 female different from all groups p<0.05, ¶ Tg3647 male different from WT male p<0.05, ‡ Tg3647 female different from WT female p<0.05, # Genotype Main Effect p<0.05 (Tg3647 different from WT at all time points).

Representative images of male (Figure 4 A and B) and female (Figure 4 C and D) Tg3647 arterioles stained with Masson Trichrome at 4 and 5.5-months old, respectively, showed an intense blue staining as well as a thickened medial layer around the vessels (Green Arrow, Figure 4C). The arteriole wall displayed dramatic thickening in the 4 and 5.5-month old Tg3647 female mice compared to all the other groups (Figure 4E). No differences in vessel wall thickening amongst male WT, female WT or male Tg3647 mice was found at any age. Further, the ratio of wall area to luminal area was increased from 4 to 5.5 months of age in Tg3647 female suggesting progressive narrowing of the arterioles (Red Arrow Figure 4 D, F). Based on such severe pulmonary arteriole pathology, we next valuated right ventricular (RV) enlargement and found it was significantly increased (Black Arrow) in female transgenic mice compared to male transgenics at 4 and 5.5 months of age and WT littermates (Figure 4G–K, Supplemental Figure 6 A and B). Further, a significant correlation between lung arteriole wall area and RV area was noted (Spearman’s Rho=0.70, p<0.0001). Assessment of liver, kidney and spleens from three male and three female 5.5-month old WT and Tg3647 was performed. In agreement with the initial necropsy study of four Tg3647 females, necropsy of WT internal organs revealed no pathology while Tg3647 mice had severe cardiopulmonary disease with mild liver and kidney congestion indicative of heart failure. Additionally, there was splenomegaly consistent with prior reports of Tg3647 mice [8, 9, 15, 21].

Figure 4. Female Tg3647 mice have significant vascular pulmonary pathology concomitant with right ventricular enlargement.

Representative Masson Trichrome stained lung sections of 4 and 5.5-month old male (A, B) and female (C, D) Tg3647 show pronounced arteriole thickening (Green Arrows) in the female transgenics (E). Furthermore, the arterioles begin to close off (Red Arrows) at 5.5 months as demonstrated by the significant increase in the ratio of arteriole wall area to luminal area compared to 4 months of age (F). Cardiac histopathology of 4 and 5.5-month old male (G, H) and female (I, J) Tg3647 reveals significant enlargement of the right ventricle (Black Arrows) in the female Tg3647 at both 4 and 5.5 (K) compared to their male counterparts. A significant correlation was also found between the arteriole wall area and right ventricular wall area (L, Spearman’s Rho=0.7, p<0.0001). Statistical results: † Tg3647 female different from TNF male p<0.05, § Tg3647 female different from all groups p<0.05, ¶ Tg3647 male different from WT male p<0.05, ‡ Tg3647 female different from WT female p<0.05, # Genotype Main Effect p<0.05 (Tg3647 different from WT at all time points), *p<0.05.

CD11b+/CD11c+ Double Positive Cells are Increased in Tg3647 Lungs

In order to evaluate cellular and molecular characteristics related to the cardiopulmonary histological changes, we performed flow cytometry and cytokine analysis on our cohorts of mice. Total cell numbers of enzymatically digested lungs were increased in transgenic mice at all time points with total cell numbers peaking at 4 months of age in Tg3647 female mice (Figure 5A). All immune cell populations studied were also increased at all time points compared to their WT counterparts (Figure 5B–F). Interestingly, when percent populations were calculated, the percentages of CD3 and CD19 cells in the lung were not elevated (Figure 5G, H), yet there was a dramatic increase in CD11b/CD11c double positive cells by percentage for both male and female Tg3647 mice compared to their WT counterparts at all time points (Figure 5K). Notably, CD11b positive cell percentages were significantly increased at all ages in both male and female Tg3647 mice compared to WT controls (Figure 5I). CD11c positive percentages were increased at 3 months in the transgenic mice, but were not different at 4 months compared to WT animals and increased again at 5.5 months of age (Figure 5K). Due to the lack of consistency between cell counts and cell area measured by histomorphometry at 5.5 months of age (Figure 5A vs Figure 2C), lung slides were stained with a TUNEL assay to assess apoptosis. Indeed, a significant increase in apoptosis at both 4 and 5.5 months was observed in Tg3647 female lung tissue compared to all other groups and an increase from 4 to 5.5 months of age was noted, suggesting that the lower cell counts determined by flow cytometry vs. histology were due to substantial cell loss ex vivo.

Figure 5. CD11b/CD11c double positive cells are significantly increased in Tg3647 mice.

Total live cells were significantly more abundant in the Tg3647 lungs at all time points compared to WT littermates (A). These differences in cell counts were maintained between WT and TNF for all cell populations over time identified by flow cytometry (B-F). Percent populations for CD3 and CD19 showed no significant difference (G and H). However, CD11b%, CD11c% and CD11b/CD11c% were increased in the Tg3647 mice compared to their WT counterparts (I, J, and K). TUNEL stain reveals a significant amount of cell death in the female Tg3647 at 4 and 5.5 months of age as well as an increase from 4 to 5.5 months of age (L). Statistical results: † Tg3647 female different from TNF male p<0.05, § Tg3647 female different from all groups p<0.05, ¶ Tg3647 male different from WT male p<0.05, ‡ Tg3647 female different from WT female p<0.05, # Genotype Main Effect p<0.05 (Tg3647 different from WT at all time points), ***p<0.001.

Because the Tg3647 mouse line contains a copy of the human TNF (hTNF) gene as well as its own mouse TNF gene, we measured both human and mouse TNF levels. Cytokine analysis revealed a significant increase in hTNF levels in the serum of all Tg3647 mice at all time points (Figure 6A) compared to WT. IL-9 and IL-10 levels were increased in the Tg3647 female mice only (Figure 6B–E), while IL-17 was increased in the both WT and Tg 3647 female mice compared to males. IP-10, MCP-1, LIF, and KC were all increased in both sexes of the Tg3647 mice (Figure 6F–J).

Figure 6. Serum cytokines and chemokines are increased in the Tg3647 mice compared to their WT littermates.

Human TNF levels were elevated in both sexes of the Tg3647 compared to WT littermates (A). However, serum concentrations of mouse TNF at 5.5 months of age are similar for all groups (B). IL-9 and IL-10 were significantly more abundant in Tg3647 female serum than their male counterparts, while IL-6 serum concentration was comparable in all (C-E). Interestingly, IL-17 was significantly increased in the female sex compared to males independent of genotype (F). IP-10, MCP-1, LIF and KC were all significantly increased in the Tg3647 mice independent of sex (G-J). Statistical results: † Tg3647 female different from TNF male p<0.05, § Tg3647 female different from all groups p<0.05, ¶ Tg3647 male different from WT male p<0.05, ‡ Tg3647 female different from WT female p<0.05, # Genotype Main Effect p<0.05 (Tg3647 vs WT), & Sex Main Effect (Male vs Female) p<0.05.

Discussion

Rheumatoid arthritis is a systemic autoimmune disease and there are many established animal models to study joint and extra-articular RA-like disease [11], but to our knowledge, this report is the first to longitudinally characterize sexual dimorphism in a spontaneous murine model of chronic inflammatory-erosive arthritis with concomitant cardiopulmonary pathology. In female mice, the progressive cardiopulmonary disease becomes lethal in the first several months, and explains the mechanism of early mortality of females in this well-established model of RA. We also describe significant sexual dimorphism in: muscle dysfunction, arthritic onset, timing of joint-draining lymph node collapse, interstitial lung disease and pulmonary arteriole thickening.

Extra-articular manifestations of RA, in particular lung disease, are a common occurrence. Recent estimates report that up to 10% of the RA population experience pulmonary involvement [22–24]. Furthermore, patients diagnosed with interstitial lung disease (ILD) have poor median survival expectancy with some studies reporting as low as 2.6 years median survival after diagnosis, and growing evidence suggests an increased mortality incidence in female patients [23, 25]. In a recent review from our group on RA-ILD, we discuss the lack of consensus regarding the relationship between sex and RA-ILD morbidity and mortality [26]. While historically men with RA have been presumed to have increased all-cause mortality compared to women, more recent data suggests that women may be at higher risk of death from RA, in particular related to ILD. In a large retrospective study of more than 2 million decedent records from the National Center for Health Statistics in the US, Olson et al found that in decedents with RA-ILD, mortality rates on average were increased in women compared to men for all age groups [23]. Although the authors did not speculate on the etiologies of this increase in women, we posit that the sexual dimorphism observed in the Tg3647 mouse may inform this discussion. Furthermore, a recent study reported that sexual dimorphism is present in subtypes of RA-ILD (i.e. non-specific interstitial pneumonia vs usual interstitial pneumonia, discussed further below), which may explain some of the incongruous findings in clinical studies of sex differences and mortality of RA-ILD [27].

The etiopathogenesis of RA-ILD is poorly understood, and diagnosis of lung disease is primarily based on imaging modalities. The predominant types of RA associated intestinal lung disease are a fibrotic type, usual interstitial pneumonia (UIP), and a cellular infiltrate type, non-specific interstitial pneumonia (NSIP) [27]. Frequently, however, both imaging and histology via lung biopsies have shown that some patients may have characteristics of both subtypes [24], suggesting a spectrum of disease rather than distinctly different types. In our mouse model, we see an interstitial cellular infiltrate similar to NSIP. While the fibrotic UIP subtype is often thought to have a worse prognosis, recent evidence suggests that both types have similar mortality [28, 29].

A notable phenomenon of the phenotype in the Tg3647 model is the pulmonary arteriole thickening that likely leads to right ventricular hypertrophy (Figure 4), and mortality with dramatic cardiopulmonary pathology. Clinically, RA associated pulmonary hypertension is rare but one of the predominant causes is interstitial lung disease, as seen in this model [30]. Other causes include vasculitis and thromboembolic disease, neither of which were identified in our model. Interestingly, development of pulmonary hypertension is much more common in other connective tissue diseases, such as systemic sclerosis, mixed connective tissue disease and lupus [31, 32]; furthermore, the role of TNF in the development of pulmonary vasculopathy has been intensely studied and has been described in animal models of TNF overexpression [33, 34]. However, to date, there is no preclinical model to recapitulate cellular ILD related effects on vasculopathy and mortality outcomes. Therefore, use of this sexually dimorphic Tg3647 model in future research studies are warranted to better understand the pathophysiologic processes that may be clinically relevant to connective tissue diseases.

Importantly, others have seen similar sexual dimorphism in acute arthritis animal models of RA. Keith and colleagues investigated the role of testosterone on arthritis and lung disease in the Zymosan induced SKG mouse model [4]. They found increased incidence and severity of arthritis in female SKG mice injected with Zymosan compared to male mice, and orchiectomized male SKG mice displayed similar incidence and severity as the females. Furthermore, they identified an inflammatory, but not fibrotic, lung disease in both the female SKG and orchiectomized males, but not male SKG mice only injected with Zymosan. Interestingly, arthritic disease worsened in ovariectomized SKG mice, and arthritis was ameliorated in sham animals treated with estrogen [35]. Studies of the collagen induced arthritis (CIA) model demonstrated increased incidence in male mice after immunization, and male castration exacerbated disease [36, 37]. When either testosterone or estrogen was exogenously delivered to immunized animals of either sex, they suppressed disease [38–40]. These data indicate that sex hormones can play both protective and accelerating roles in autoimmune disease depending on the source of inflammation and the type of response initiated by the immune system. Thus, studies modulating the sex hormones in Tg3647 mice are necessary to better understand their interactions within the immune system, the sexual dimorphism and the multisystem pathologies.

Contrasting effects of sex hormones on autoimmunity are also seen in RA patients. Women are significantly more likely to have RA than men, and there is a modest reduction in disease activity during pregnancy, with women who have moderate or severe disease seeing the largest effect [41]. There is also a significant risk of flare postpartum. Yet women with RA who take oral contraceptives do not show a clear reduction in relative risk (RR: 0.88 95% CI:0.751–1.03) [42]. Androgen replacement therapy may be effective at improving disease outcomes, especially in men and women with low testosterone levels [43, 44]. Overall, several factors including serum versus tissue concentration, timing and clinical features may dictate the effect of sex hormones on arthritis.

Recently, Ntari et al described a strictly cardiovascular disease in the 197 line of TNF-Tg mice [45], which is a more aggressive disease model than the 3647 TNF-Tg line described here [10]. These investigators found left sided aortic and mitral valve thickening with significant fibrosis with no evidence of lung inflammation or right heart pathology. This effect was driven by TNF receptor signaling dependent on valvular interstitial cells. In our model, however, we found no evidence of left heart pathology in Tg3647 mice (Figure 4 and Supplemental Figure 7). We did observe significant collagen deposition in the interstitium of the myocardium of the right ventricle, suggesting that TNF signaling may be conserved in both processes (Supplemental Figure 6C–F). These difference between the two prominent TNF-Tg mouse lines are likely due to the number of transgenes inserted, the age of onset of pathology and the relative amounts of circulating huTNF. As mentioned earlier, the Tg197 has earlier onset arthritis and cardiovascular disease than the Tg3647 and no sexual dimorphism has been reported [8, 9]. In addition, the pulmonary vasculopathy is primarily restricted to the females in the Tg3647 line, suggesting that TNF may be interacting significantly with the female hormones whereas the Tg197 develop disease before sexual maturity. This combination of factors likely account for the differences in pathology between the two primary TNF-Tg mouse models.

Interestingly, when we performed flow cytometry on the lungs of Tg3647 mice, CD11b/CD11c double positive cells were significantly increased, both in total numbers and by percent, in the male and female mice at all time points (Figure 5). This cell population is likely to be of dendritic cell origin [46], which has been suggested to play a key role in immunopathology of lung disease [47], and is likely the major component of the interstitial inflammatory infiltrate into the lung. Further, ectopic lymphoid follicles were seen in histology (Supplemental Figure 6) and total B cell numbers were increased by flow cytometry. The presence of ectopic lymphoid tissue in our transgenic is similar to the inducible Bronchus Associated Lymphoid tissue (iBALT) seen in patients with pulmonary complications of RA and Sjogren’s syndrome [48]. Consistent with the idea that iBALT supports local production of auto-antibodies, antibodies against vimentin and H2B detected in the sera of the Tg3647 line may be produced in the lung [49]. A limitation of our flow cytometry data was that the total cell counts for female Tg3647 lung tissue did not correlate with the histologic cell counts (Figure 5A vs Figure 2C). One explanation is that the cells in these lungs were either primed to undergo programmed cell death or were currently undergoing apoptosis. To confirm this hypothesis, we stained lung tissue from all cohorts with a TUNEL assay and found a dramatic increase in TUNEL+ cells in the Tg3647 females from 4 months to 5.5 months (~2-fold increase). This critical and normal process observed in both inflammation and wound healing was likely limiting the number of live cells isolated and processed for flow cytometry at the later stages of disease [50].

Cytokines analysis of the Tg3647 mice found that serum IL-9 and IL-10 were increased in female Tg3647 mice, while IP-10, MCP1, LIF and KC were increased in Tg3647 mice regardless of sex (Figure 6). IL-9 is significantly increased in the sera of patients with interstitial lung inflammation associated with connective tissues disorders and is known to enhance recruitment of B-cells while reducing the incidence of interstitial fibrosis [51, 52]. Thus, it is feasible that IL-9 may be playing a role in the lack of interstitial fibrosis development in this model. IL-10 is a known anti-inflammatory cytokine and is produced to dampen the immune response in inflammatory settings. Therefore, increased IL-10 levels are likely a compensatory mechanism to counter the severe inflammation noted at 5.5 months in female Tg3647 mice. IP-10 (CXCL10), MCP1 and KC (CXCL1) are all potent chemoattractant for leukocytes and consistent with this, we find a dramatic abundance of leukocytes in the interstitium of all Tg3647 mice. In summary, serum cytokine analysis complements the histopathology and flow cytometry results to described lymphoid and myeloid cell migration and proliferation, with activated compensatory anti-inflammatory mechanisms in the Tg3647 mice.

In conclusion, we longitudinally phenotyped the specific sexual dimorphisms of the pathologies within the musculoskeletal and cardiopulmonary systems of the line TNF-Tg 3647. We found that the early mortality in the female mice is due to heart failure secondary to severe pulmonary vasculopathy. This mouse model, most commonly studied for its inflammatory-erosive arthritis, contains additional musculoskeletal and cardiopulmonary pathologies that can be studied to better understand the role of sex in TNF-mediated diseases, and specifically on these affected systems. Elucidating these mechanisms may have broad impact on the extra-articular pathologies in RA, systemic sclerosis and other connective tissue diseases.