ALPN-101 (Acazicolcept) a Dual ICOS/CD28 Antagonist, Demonstrates Ecacy in Systemic Sclerosis Preclinical Mouse Models

Uncontrolled immune response with T cell activation has a key role in the pathogenesis of systemic sclerosis (SSc), a disorder that is characterised by generalized brosis affecting particularly the lungs and skin. Co-stimulatory molecules are key players during immune activation, and recent evidence supports a role of CD28 and ICOS in the development of brosis. We herein investigated the ecacy of ALPN-101 (acazicolcept), a dual ICOS/CD28 antagonist, in two complementary SSc-related mouse models recapitulating skin brosis, interstitial lung disease, and pulmonary hypertension. Fra-2 Tg mice. Moreover, ALPN-101 reduced CD69 and PD-1 expression on CD4+ T cells from the spleen and the lung. Target engagement by ALPN-101 was demonstrated by blockade of CD28 and ICOS detection by ow cytometry in treated mice. and anti-OX40L antibody 40). Compared to targeted therapies such as the pan-PPAR agonist IVA337, we observed similar levels of lung brosis reduction in treated Fra-2 transgenic mice Interestingly, our results showed that CD69 and PD-1 expression on CD4 + T cells was positively correlated with lung collagen content, supporting a link between T cell activation and brosis development in the Fra-2 transgenic model. Our results are aligned with previous studies investigating co-stimulation blockade in SSc mouse models. Indeed, a decrease of dermal brosis and inammation after ICOS blockade in GvHD-SSc mice (28) or after intradermal bleomycin injections in ICOS −/− mice compared to WT mice (29) was observed. Other costimulation pathways blockade such as CD28-CD80/CD86 and OX40/OX40-L have previously demonstrated decreased pulmonary and dermal brosis in SSc mouse models 40, 42).


Introduction
Systemic sclerosis (SSc) is a rare autoimmune rheumatic disease characterized by vasculopathy and dysregulation of the immune response, and extensive brosis of skin and internal organs (1). This leads to increased morbidity and mortality of SSc patients mainly due to cardiovascular and pulmonary complications (2). T cells are a major component of SSc pathophysiology as indicated by their early recruitment in SSc skin (3). Several studies have shown the contribution of Th2, Th17, Th22, Tfh, and CD8 + subsets to in ammation in blood and skin of SSc patients (4). Early vascular and immune interactions are supported by the recent ndings showing that endothelial cells expressing HLA-DR are targeted by cytotoxic CD4 + cells, leading to their apoptosis and likely remodelling in affected SSc tissue (5). T cell activation, proliferation, and differentiation are based on an appropriate interaction between T cell co-stimulation molecules and their receptors on antigen-presenting cells (APC). Co-stimulation blockade in several SSc murine models has shown to mitigate brosis, linking T cell activation and brosis/remodelling development (6).
CD28 and inducible T cell costimulator (ICOS) are closely related T cell costimulatory molecules within the immunoglobulin superfamily that bind, respectively, the ligands CD80 and CD86, and ICOS ligand (ICOSL), and play partially overlapping roles in immunity (7). Signalling through CD28 and ICOS leads to T cell cytoskeletal remodelling, production of cytokines, enhanced survival, and differentiation (8, 9).
CD28 and ICOS also cooperate in lung mucosa to induce differentiation of Th2 effector cells (10). The concept of interfering with T cell costimulation to treat autoimmune diseases has been clinically validated with abatacept (CTLA-4-Ig), an approved CD28 pathway inhibitor for rheumatoid arthritis, juvenile idiopathic arthritis, and psoriatic arthritis.
The CD28 pathway inhibitor abatacept was evaluated in a phase II trial (ASSET) in early diffuse cutaneous SSc (dcSSc). Although abatacept was well-tolerated in the ASSET trial, patients treated with abatacept did not experience signi cantly greater improvements of the modi ed Rodnan Skin Score (mRSS) than those administered placebo, though some improvements in secondary outcome measures were observed in the abatacept arm (11). These results suggest that CD28 pathway inhibition alone is insu cient to signi cantly impact skin disease in dcSSc patients.
ICOS is not expressed in naïve T cells but is rapidly upregulated after activation and may represent a key pathogenic pathway unaddressed by CD28 antagonism. ICOS appears particularly important for the function of several activated and/or effector T cell subsets, including differentiated types 1, 2, and 17, as well as follicular helper T cells (TFH) (12). Indeed, activated T cells often downregulate CD28 and/or become less dependent on CD28 costimulation, and CD28-negative T cells accumulate in various in ammatory diseases, correlating with disease activity and lack of responsiveness to abatacept (13)(14)(15)(16)(17)(18)(19). In contrast, ICOS upregulation correlates with disease activity in several in ammatory diseases (20)(21)(22)(23)(24), and in preliminary studies, the anti-ICOSL mAb prezalumab (AMG-557) demonstrated some bene cial activity on the arthritis of systemic lupus erythematosus (NCT04058028 ; (25)), as well as on overall disease activity in Sjögren's syndrome (NCT02334306). However, at present no ICOS pathway antagonists have been approved for therapeutic use.
Preliminary data in SSc patients have demonstrated an increase of soluble ICOS in the sera of patients with diffuse cutaneous SSc (26, 27) and of ICOS + Tfh-like cells in their skin (28). Studies in SSc mouse models challenged with bleomycin indicated that ICOS-de cient mice were protected from skin and lung brosis (29). In a GVHD model that shares some similarities with SSc, compelling data have revealed a decrease in dermal in ammation and brosis after anti-ICOS antibody administration (28). Taken together, these data suggest a potential role of ICOS in in ammation-driven lung and skin brosis.
We hypothesized that a dual-reactive molecule that blocks both pathways, ICOS together with CD28 may be of interest in immune-related diseases; this general approach has since been shown to abrogate ongoing germinal centre reactions during an immune response (30). The blockade of CD28 and ICOS in an acute GvHD mouse model by the novel dual CD28/ICOS antagonist (ALPN-101/acazicolcept) led to improved survival in ALPN-101-treated mice compared to mice receiving a CD28-CD80/CD86 pathway antagonist (belatacept; CTLA-4-Ig) only (31). These results suggest that co-targeting ICOS and CD28 is a relevant strategy to suppress autoimmune responses (31). Therefore, we herein evaluated the therapeutic effect of ALPN-101 on immune responses and related brosis in two complementary mouse models mimicking the severe organ damage observed in SSc patients.

Materials And Methods
Animals 6-week-old female BALB/c mice were purchased from Janvier Laboratory (Le Genest Saint Isle, France) and experiments were conducted in a conventional facility (C75-14-05). Transgenic female Fra-2 (B6.Cg-Tg(H2-K-Fosl2,EGFP)13Wag) mice were bred in a SPF facility (C75-14-02). All mice were housed in ventilated cages with sterile food and water ad libitum. Animals received humane care in compliance with the guidelines implemented at our institution (INSERM and University Paris Descartes).
ALPN-101 molecule and pharmacological treatment ALPN-101 (acazicolcept; ICOSL vIgD-Fc), provided by Alpine Immune Sciences (AIS) (Seattle, WA), is a dual human ICOS/CD28 inhibitor Fc fusion protein (31). ALPN-101 (produced at KBI Biopharma, Durham NC) and Fc control protein (produced at AIS) were diluted in PBS and injected intraperitoneally twice a week at molar-matched doses of 400 µg/mouse and 267 µg/mouse, respectively. The mouse dosing regimen was identi ed from prior mouse pharmacokinetic/pharmacodynamic studies as one that provided adequate exposure and disease modifying activity in multiple mouse models of autoimmunity and in ammation. However, this dosing regimen would not be used directly to predict human regimens due to species-and disease-related differences in multiple factors including target abundance, binding, and clearance.

HOCL induction of dermal brosis and ALPN-101 treatment
Dermal brosis was induced in six-week-old BALB/c mice according to the protocol described by Servettaz et al (32). A total of 400 µL hypochlorous acid (HOCl) solution was prepared extemporaneously by adding NaClO (9.6% as active chlorine) to KH 2 PO 4 solution (100 mM, pH: 6.2), usually using a 1:100 ratio. The correct amount of NaClO was adjusted to obtain the desired HOCl concentration, de ned by the absorbance of the mixture at 292 nm (optical density between 0.7 and 0.9). 200 µL of HOCl solution was injected intradermally into each shaved ank of the mice using a 27-gauge needle, 5 days a week for 6 weeks. Control mice were injected intradermally with 200 µL of sterilized phosphate-buffer saline (PBS) into each shaved ank. 100 µL of ALPN-101 or Fc control dosing solutions were injected intraperitoneally twice a week during the 6 weeks of HOCl-treatment. Mice were divided into the following groups : PBS (n=6), HOCL + Fc control (n=8), and HOCL + ALPN-101 (n=8). Mice were euthanized by cervical dislocation after 6 weeks of treatment (Supplementary Figure 1). This experiment has been carried out once.

Fra-2 transgenic mice and ALPN-101 treatment
Transgenic mice expressing the Fra-2 transgene under the control of ubiquitous major histocompatibility complex class I antigen H-2K b promoter develop microangiopathy, systemic in ammation, lung brosis, and pulmonary hypertension (33). These features follow a similar temporal sequence as observed in human SSc. In the lungs, perivascular in ammatory in ltrates and vascular remodeling appear at the 12 th week of age and are followed by brosis development at 15 th week of age (34). Fra-2 transgenic mice display severe vascular remodeling of pulmonary arteries leading to their intimal thickening and, in the worst case, to obliteration of vessels (35). Two groups of Fra-2 transgenic female mice were treated starting at 12 weeks of age with intraperitoneal injections of ALPN-101 (n=11) or Fc control (n=8) twice a week, for a total of 6 weeks. Mice were euthanized by exsanguination after right ventricular systolic pressure (RVSP) measurement at 18 weeks of age (Supplementary Figure 2). This experiment has been carried out twice.

ALPN-101 serum measurement
The concentration of ALPN-101 was measured in serum samples collected 24 hours after the 8 th or 13 th dose in the HOCL model, or after the 10 th and 13 th dose in the Fra-2 Tg model, using an ELISA method developed at Alpine Immune Sciences. ALPN-101 was captured by Fc-speci c donkey anti-human IgG antibody (Jackson ImmunoResearch), immobilized onto a 96-well microtiter plate and detected with F(ab') 2 fragment, Fc-speci c donkey anti-huIgG:HRP (Jackson ImmunoResearch). A calibration curve was generated for each assay plate using SoftMax Pro data acquisition and analysis software (version 7.1, Molecular Devices).
Clinical follow-up of Fra-2 mice Fra-2 transgenic mice developed a disease phenotype requiring their clinical follow-up. Monitoring included weighing the mice once a week for the duration of the experiment. All the mice were scored individually using body weight change and observation of their physical appearance and behavior. Mice received a clinical score of 0 to 3, with 0 = normal ; 1 = weight loss <10%, lack of grooming and behavior minor modi cations ; 2 = weight loss between 10-15%, alopecia and skin lesions, reduced mobility, Raynaud's syndrome ; 3 = weight loss > 20%, ru ed fur, hunched posture, lethargy. If mice reached a clinical score of 3 before the end of the experiment, they were euthanized to respect the 3R rule.
Skin thickness measurement of HOCL-treated mice Skin thickness (expressed in millimeters) was assessed using a caliper to measure the dermal thickness of the shaved backs of the mice. The measurement was performed once a week until the end of the experiment.

Collagen measurement
Collagen content was measured in a 3-mm punch from the back skin of HOCL-treated mice or from lung biopsies (right lobes) of Fra-2 mice using Sircol® soluble collagen assay (Biocolor, UK) according to the manufacturer's instructions. Collagen content was determined from the slope of the standard curve calculated using known collagen concentrations.
Analysis of the immunostaining was performed using the Lamina Multilabel Slide Scanner (Perkin Elmer, USA). Slide staining analysis was performed with CaseViewer software (version 2.4).
Histopathologic assessment of dermal brosis in HOCL-treated mice and brosing alveolitis in the Fra-2 model Fixed 6-mm skin punch biopsies from HOCL-treated mice or left lung from Fra-2 mice were embedded in para n. A 4-µm thick tissue section was stained with hematoxylin, eosin, and saffron. Slides were scanned with the Lamina Multilabel Slide Scanner. For HOCL skin sections, dermal thickness was evaluated at 100-fold magni cation by measuring the distance between the epidermal-dermal junction and the dermal-subcutaneous fat junction at ve sites on skin sections by two independent blinded examiners with the CaseViewer software (version 2.4). The mean of the 10 values obtained by the two examiners was calculated for each skin section. For Fra-2 Tg lung sections, the severity of brosing alveolitis was semi-quantitatively assessed by examining the entire slide, by two examiners blinded to the treatment. The grading criteria were as follows: 0 = normal lung ; 1 = Minimal brous thickening of alveolar or bronchioalveolar walls ; 2-3 = moderate thickening of walls without obvious damage to lung architecture ; 4-5 = Increased brosis with de nite damage to lung structure and formation of brous bands or small brous masses ; 6-7 = Severe distortion of structure and large brous areas and 8 = Total brous obliteration (36).
Nonlinear microscopy and second harmony generation (SHG) processing A 2-photon Leica SP8 DIVE FLIM (Leica Microsystems GmbH, Wetzlar, Germany) was used for lung and skin tissue imaging. Two lasers at 1040 and 880 nm wavelength were used to generate second harmonic (SHG) and two-photon-excited uorescence (TPEF) signals, collected by a Leica Microsystems HCX IRAPO 25×/0.95 W objective and two external detectors. Microscopy was performed on 16 µm-thick blank blades of sliced lungs or skin. Five samples of each slice were taken. The SHG score was established by comparing the area occupied by the collagen relative to the sample surface. Image processing and analysis (thresholding and SHG scoring) were performed using ImageJ homemade routine as previously described (37).
Right ventricular systolic pressure (RVSP) measurement in Fra-2 mice RVSP was assessed in unventilated mice under iso urane anesthesia (1.5-2.5%, 2L O 2 /min) using a closed chest technique by introducing a catheter (1.4-F catheter; Millar Instruments Inc., Houston, TX) into the jugular vein and directing it to the right ventricle. After RVSP measurement, blood was collected by direct cardiac puncture leading to mouse sacri ce. The heart and lungs were removed and ushed with 5 mL of buffered saline at 37°C. The left lung was xed in paraformaldehyde 4%. For 10 Fra-2 Tg mice (4 Fc control-and 6 ALPN-101-treated), one lobe of the right lung was collected to perform FACS analysis and other lobes were immediately snap-frozen in liquid nitrogen and kept at -80°C.
Spleen and lung cell isolation for ow cytometry staining Flow cytometry staining was performed on 4 spleen/lung Fc control-treated mice and on 6 spleen/lung ALPN-101-treated mice. Spleens were collected and crushed on a 70 µm cell strainer. Red cells were removed with ACK Lysing Buffer (Thermo sher). 1x10 6 cells were collected for ow cytometry staining. One lung lobe was cut in small pieces and incubated in PBS 10% FBS + Collagenase II (1mg/mL, StemCell Technologies) and DNase I (0.1mg/mL, StemCell Technologies) for 1 hour at 37°C. After mechanical dissociation with vortexing, cell suspensions were passed through a 70 µm cell strainer. Red blood cells were removed with ACK Lysing Buffer (ThermoFisher). A Percoll (Sigma) density gradient was generated by resuspending cells in a 40% Percoll solution and adding an 80% Percoll solution below the 40% solution. The cell ring was collected, and the total lung cell suspension was used for ow cytometry staining. Spleen

ICOS measurement in human serums
ICOS protein was quanti ed by ELISA in the serum of 161 patients with SSc and 38 healthy age-and sexmatched volunteers using the Human ICOS (CD278) ELISA Kit (ThermoFisher Scienti c™) according to the manufacturer's instructions.

Statistics
All data analysis were performed using GraphPad Prism 9 Software. Human data were presented as mean with standard deviation (SD) and analyzed with Student's t-test. Mouse data were presented as median with ranges and analyzed by Mann-Whitney Test. Correlation data were analyzed with Spearman's correlation test. A p value of less than 0.05 was considered statistically signi cant.

ICOS expression is increased in serum and skin of SSc patients
We evaluated ICOS expression in the serum of SSc patients (n=161) and healthy controls (n=35). We observed a higher concentration in SSc patients compared to controls: 20.10 ng/mL +/-31.31 in SSc versus 7.97 ng/mL +/-6.28 in controls (p=0.024) ( Figure 1A). After strati cation on skin subsets, we observed no difference between diffuse and limited cutaneous SSc patients: Diffuse 20.91 ng/mL +/-29.53 (n=68) versus Limited : 19.5ng/mL +/-32.70 (n=93) ( Figure 1B). The sub-group de ned by the presence of interstitial lung disease (n=51) was not associated with higher ICOS concentration as compared to patients free of ILD (n=110) ( Figure 1B). No other SSc subset including other major organ involvement, disease duration or auto-antibodies, was associated with different serum concentrations.
We next investigated CD3+ ICOS+ T cells in healthy controls and SSc skin taken from dcSSc patients. We observed few and isolated CD3+ ICOS+ T cells in control skin whereas aggregates of CD3+ ICOS+ T cells were readily detected in lesional SSc skin ( Figure 1C).
Evaluation of ALPN-101 E cacy in HOCL-Induced Dermal Fibrosis ALPN-101 prevents HOCL-induced dermal brosis ALPN-101/HOCL-treated mice had similar body weight changes as observed for the Fc Control/HOCL and PBS group ( Figure 2A). As HOCL injections induce skin thickening, we measured dorsal skin folds with a caliper from week 1 to week 6. After 6 weeks of treatment, skin fold thickness was increased by 1.5-fold in HOCL/Fc control-treated mice compared to PBS-treated mice (p=0.0007). ALPN-101 treatment signi cantly decreased the skin fold thickness by 17.5% compared to HOCL/Fc control-treated mice (p=0.0012) ( Figure 2B).
Skin sections from HOCL/Fc control mice were characterized by marked skin thickening as shown in Figure 2C. Dermal thickness was 1.7-fold-increased in HOCL/Fc control-treated mice compared to PBStreated mice (p=0.0007). A signi cant decrease of dermal thickness by 25.5% was observed in HOCL/ALPN-101-treated mice compared to HOCL/Fc control treated-mice (p<0.001) ( Figure 2C).
Skin collagen content was 1.3-fold higher in HOCL/Fc control mice compared to PBS-treated mice (p=0.003). A signi cant reduction of collagen content by 20.6% was observed in HOCL/ALPN-101-treated mice compared to HOCL/Fc control-treated mice ( Figure 2D).  Figure 3A and B).
Evaluation of ALPN-101 e cacy in Fra-2 mouse model ALPN-101 treatment reduces clinical scores in Fra-2 mice In general, the body weight of mice receiving ALPN-101 was maintained throughout the experiment compared to mice treated with Fc control that lost body weight with age, though the difference between the groups was not statistically signi cant ( Figure 4A, left). Clinical scores evaluating weight loss, coat appearance, and mouse behaviour decreased by 78.4% (p=0.008) and 72.2% (p=0.012), respectively, at the 5 th and 6 th week in ALPN-101-treated Fra2 Tg mice vs. Fc control-treated Fra2 Tg mice ( Figure 4A, right).

ALPN-101 treatment alleviates lung brosis and pulmonary hypertension in Fra-2 mice
ALPN-101 treatment decreased collagen content signi cantly in lungs from Fra-2 Tg mice, by 35.2% (p=0.005) compared to Fc control ( Figure 4B). Lung sections of Fc control-treated Fra-2 Tg mice were characterized by large patchy areas of in ammatory in ltrate and collagen deposition ( Figure 4C). The histological Ashcroft score of brosis was signi cantly reduced by 33.3% (p= 0.032) in ALPN-101-treated Fra2 Tg mice compared to Fc control-treated Fra2 Tg mice ( Figure 4C). SHG microscopy showed an increase of collagen bers around lung vessels ( Figure 4D) in Fc control-treated Fra2 Tg mice. Decreased brillar collagen deposition by 47% (p=0.06) was observed in ALPN-101-treated mice compared to Fc control-treated Fra2 Tg mice ( Figure 4D).
Regarding pulmonary hypertension (PH), a signi cant reduction (20.3%, p=0.019) of right ventricular systolic pressure (RVSP) was observed in Fra-2 Tg mice treated with ALPN-101 compared to Fra2 Tg mice that received Fc control treatment.

ALPN-101 reduces T cell response in spleen and lungs of Fra-2 mice
To evaluate the effects of ALPN-101 on T cell responses, we performed ow cytometry analysis by gating on CD4+ and CD8+ populations isolated from the spleen and lungs of treated Fra-2 Tg mice (Supplementary Figure 3). Treatment with ALPN-101 signi cantly reduced the percentage of CD4+ cells by 11.9% in the spleen (p=0.0381) and by 27.6% in the lungs (p=0.009) compared to Fc control-treated Fra2 Tg mice ( Figure 5A). No signi cant changes in percentages of CD8+ cells were observed between Fc control-and ALPN-101-treated Fra2 Tg mice ( Figure 5A). We next investigated the proportions of effector memory T cells (TEM), naïve T cells (T Naïve), and central memory T cells (TCM) based on their differential expression of CD62L and CD44. A signi cant decrease of CD4+ TEM cells by 43.9 % in spleen and by 23.8% in lungs (p=0.009) in ALPN-101-treated Fra2 Tg mice was observed compared to Fc controltreated Fra2 Tg mice. The frequency of CD4+ T naive cells was signi cantly increased by 2.6 times in spleen and by 4 times in lungs (p=0.009) in ALPN-101-treated Fra2-Tg mice compared to Fc controltreated Fra-2 Tg mice ( Figure 5B). No differences between Fc control-and ALPN-101-treated Fra2 Tg mice were observed in the proportions of CD4+ TCM, CD8+ TCM, CD8+ TEM, or CD8+ naïve T cells in spleen and lungs.
Activation of CD4+ and CD8+ T cells was assessed based on the expression of the early activation marker CD69 and the T cell exhaustion marker PD-1. The fraction of CD69-expressing cells was signi cantly reduced by 63.6% within the CD4+ subset in the spleen, (p=0.009) and by 58.2% among CD4+ cells in the lung (p=0.038) upon treatment with ALPN-101 compared to Fc control treatment ( Figure  5C). ALPN-101 treatment induced a signi cant decrease by 60% of CD69-expressing cells among CD8+ spleen cells (p=0.019), but the decrease in CD69-expressing cells within the CD8+ subset in the lung was not statistically signi cant (p=0.26), compared to Fc control-treated Fra2 Tg mice ( Figure 5C). Upon treatment with ALPN-101, a signi cant 38.2% and 43.2% reduction of PD-1-expressing cells was observed within the CD4+ subset in the spleen (p=0.0095) and in the lung (p=0.038) compared to Fc control treatment, respectively. No changes in the frequency of PD-1-expressing cells were detected within the CD8+ subset in the spleen or lung between the two groups of mice.
Interestingly, we detected a strong correlation between the lung collagen content and CD69 or PD-1 expression in lung CD4+ cells (r=0.9478, p<0.001 and r=0.8545, p=0.003, respectively) ( Figure 5D), linking immune activation and extracellular matrix production. Similar ndings were observed for CD8+ T cells ( Figure 5D).

ALPN-101 serum exposure
We observed 24 hours after ALPN-101 injection similar concentrations between the 10 th dose and the 13 th dose of ALPN-101 in Fra-2 Tg mice (10 th dose, mean ± SD :42630 ± 12112 ng/mL versus 13 th dose : 33728 ± 8591 ng/mL) and between the 8 th dose and the 13 th dose in HOCL-treated mice (8 th dose : 24065 ± 13359ng/mL versus 13 th dose : 25239 ± 12090ng/mL) ( Figure 6A). To track ALPN-101 binding to target cells, we stained cells isolated from spleen and lung with anti-human IgG Fc, which is able to detect the Fc domain of ALPN-101 (Supplementary Figure 4). A signi cant increase of anti-human IgG staining on spleen and lung CD4+ and CD8+ T cells (p=0.009) was observed in ALPN-101-treated Fra-2 mice compared to Fc control-treated Fra-2 mice ( Figure 6B) suggesting ALPN-101 was bound to the majority of T cells. Since ALPN-101 blocks detection of its targets CD28 and ICOS, we rst assessed CD28 expression on splenic and lung T cells by ow cytometry, as a method to track target occupancy (Supplementary Figure 4). We observed a reduced detection of CD28 on spleen CD4+ T cells by 99.7% and spleen CD8 T cells by 98.9 % (p=0.009) in ALPN-101-treated Fra2 mice compared to Fc controltreated Fra2 Tg mice. Detection of CD28 was signi cantly decreased by 66% and by 82.4%, respectively, in lung CD4+ T cells and CD8+ T cells (p=0.0095) isolated from ALPN-101-treated Fra2 Tg mice ( Figure   6C). We next analysed ICOS expression on lung and spleen T cells from ALPN-101-and Fc control-treated Fra-2 Tg mice. Detection of ICOS in spleen cells was signi cantly decreased by 98.2% on CD4+ (p=0.005) and by 81.2% on CD8+ cells (p=0.009) from ALPN-101-treated Fra2 Tg mice compared to Fc controltreated Fra2 Tg mice. Similar to the spleen results, detection of lung ICOS expression was signi cantly reduced by 99.7% on CD4+ (p=0.009) and by 88.1% on CD8+ (p=0.

009) T cells in ALPN-101-treated Fra2
Tg mice ( Figure 6C). These results demonstrated target engagement of ALPN-101 in the spleen and lungs of Fra-2 Tg mice.

Discussion
We herein showed the overexpression of ICOS in SSc patients and demonstrated the e cacy of ALPN-101, a dual CD28/ICOS antagonist, in two complementary mouse models mimicking severe features of SSc patients.
The HOCL-induced dermal brosis model, based on induction of oxidative stress by hypochlorite, is characterized by dermal in ammation, broblast activation, and collagen production (32) as observed in SSc patients (38). We observed a decrease of dermal thickness, collagen content, myo broblast number, and in ammatory in ltrate in ALPN-101-treated HOCL-induced mice. These compelling data support a bene t of ALPN-101 treatment in reducing the skin involvement in the HOCL mouse model.
The transgenic Fra-2 mice, in which immune in ltration is followed by pulmonary brosis and pulmonary hypertension (33), recapitulates several severe features affecting internal organs of SSc patients (1). Our study demonstrated that ALPN-101 treatment decreased lung brosis and collagen content, right ventricular systolic pressure (RVSP), and T cell numbers and activation in Fra-2 Tg mice. The magnitude of the effect of ALPN-101 is in line with other co-stimulation blockade therapies already studied in the Fra-2 model such as abatacept and anti-OX40L antibody (39,40). Compared to targeted therapies such as the pan-PPAR agonist IVA337, we observed similar levels of lung brosis reduction in treated Fra-2 transgenic mice (41). Interestingly, our results showed that CD69 and PD-1 expression on CD4 + T cells was positively correlated with lung collagen content, supporting a link between T cell activation and brosis development in the Fra-2 transgenic model. Our results are aligned with previous studies investigating co-stimulation blockade in SSc mouse models. Indeed, a decrease of dermal brosis and in ammation after ICOS blockade in GvHD-SSc mice (28) or after intradermal bleomycin injections in ICOS −/− mice compared to WT mice (29) was observed. Other costimulation pathways blockade such as CD28-CD80/CD86 and OX40/OX40-L have previously demonstrated decreased pulmonary and dermal brosis in SSc mouse models (39,40,42).
In humans, abatacept (CTLA-4-Ig) has been evaluated in a recent phase II study showing a trend of decreased mRSS in early diffuse cutaneous SSc patients treated with abatacept without reaching signi cance compared to placebo group (11). Interestingly, the decline in mRSS was higher in abatacepttreated patients belonging to in ammatory and normal-like skin gene expression subsets compared to the placebo group, providing support for co-stimulation blockade as a therapeutic strategy for these in ammatory patients.
Our study revealed an increase of ICOS concentration in a large set of SSc patients, extending data obtained in previous studies (26,27). Moreover, a higher number of circulating T follicular helper cells expressing ICOS was reported in SSc (43) compared to controls; such changes have also been reported in Sjögren's syndrome (44), systemic lupus erythematosus (45), and rheumatoid arthritis (22). These human data were complemented by preclinical work reporting reduced disease progression and humoral responses in lupus nephritis and collagen-induced arthritis mouse models after ICOS-L blockade (46). An anti-ICOSL antibody (AMG557) has been evaluated in patients affected by Sjögren's syndrome (NCT02334306) or by active SLE (NCT04058028) but has revealed no statistically signi cant e cacy in treated patients compared to placebo group, suggesting that inhibition of the ICOS pathway alone may be insu cient to impact disease. Altogether, the preclinical and clinical results support a role for both the ICOS and CD28 pathways in connective tissue disorders, and our data herein extend the ndings to the speci c brotic phenotype that characterises SSc.
One limitation from the study herein might be that it does not address whether the dual-speci c compound ALPN-101 may offer a bene t as compared to single therapies targeting CD28 or ICOS alone, however each single therapy has already demonstrated its relevance in SSc (28, 29,39,42)