Local administration of mesenchymal stromal cells is safe and modulates the immune compartment in ulcerative proctitis

BACKGROUND Due to their immunoregulatory and tissue regenerative features, mesenchymal stromal cells (MSCs) are a promising novel tool for the management of ulcerative proctitis (UP). Here we report on a phase IIa clinical study that evaluated the impact of local MSC therapy on UP. METHODS Thirteen refractory UP patients, with an endoscopic Mayo score (EMS) of 2 or 3, were included. Seven patients received 20–40 million allogeneic MSCs (cohort 1), while 6 patients received 40–80 million MSCs (cohort 2). Adverse events (AEs) were assessed at baseline and on weeks 2, 6, 12, and 24. Clinical, endoscopic, and biochemical parameters were assessed at baseline and on weeks 2 and 6. Furthermore, we evaluated the engraftment of MSCs, the presence of donor-specific human leukocyte antigen (HLA) antibodies (DSAs), and we determined the impact of MSC therapy on the local immune compartment. RESULTS No serious AEs were observed. The clinical Mayo score was significantly improved on weeks 2 and 6, and the EMS was significantly improved on week 6, compared with baseline. On week 6, donor MSCs were still detectable in rectal biopsies from 4 of 9 patients and DSAs against both HLA class I and class II were found. Mass cytometry showed a reduction in activated CD8+ T cells and CD16+ monocytes and an enrichment in mononuclear phagocytes and natural killer cells in biopsies after local MSC therapy. CONCLUSION Local administration of allogeneic MSCs is safe, tolerable, and feasible for treatment of refractory UP and shows encouraging signs of clinical efficacy and modulation of local immune responses. This sets the stage for larger clinical trials. TRIAL REGISTRATION EU Clinical Trials Register (EudraCT, 2017-003524-75) and the Dutch Trial Register (NTR7205). FUNDING ECCO grant 2020.

baseline. Slight inflammation in other parts of the colon was accepted with a maximum EMS of 1. 26 Patients with UP were refractory to conventional medical therapy, which means that at some point 27 during the course of the disease, patients had received rectal 5-ASA therapy and rectal 28 corticosteroid therapy for at least four weeks without an adequate response to treatment. If 29 patients were on rectal therapy, this was stopped 2 weeks before endoscopic MSC therapy and only 30 restarted after 6 weeks. If patients were on oral 5-ASA therapy or corticosteroids, the dose was kept 31 stable for 4 and 2 weeks, respectively prior to study entry and remained the same dose during the 32 first 6 weeks after MSC therapy. Patients treated with 6-mercaptopurine, methotrexate, 33 azathioprine, vedolizumab, or anti-tumor necrosis factor (TNF)-therapy should have been on 34 medication for 3 months with a stable dose for 2 months prior to study entry, and remained on the 35 same dose during the first 6 weeks of MSC therapy. Females of child-bearing age should be non-36 pregnant, non-breastfeeding, and should use adequate contraception. Patients with expected 37 hospitalization or surgery within 3 months were excluded. Additional exclusion criteria were 38 evidence of any infection needing antibiotic treatment; renal or hepatic failure; use of any 39 investigational drug within one month prior to screening or within five half-lives of the 40 investigational agent (whichever is longer); positive stool culture for enteric pathogens (salmonella, 41 shigella, yersinia, and campylobacter), positive Clostridium difficile toxin A and B, positive stool ova, 42 and parasite exam; tuberculosis or an opportunistic infection within six months prior to screening; 43 positive test for hepatitis B/C, polymerase chain reaction (PCR)-cytomegalovirus, PCR-Epstein-Barr 44 virus, or human immunodeficiency virus serology at baseline. Further exclusion criteria were active 45 malignancy in the past five years; any dysplasia in the colon in the past 5 years; or previous 46 treatment with allogeneic MSCs. 47 48 Detection of IgG HLA-antibodies in serum samples from baseline and week 6 was performed using 52 Lifecodes Life Screen Deluxe Kit (Immucor) (LMX) and further tested with single bead antigen assay 53 (SBA) Lifecodes LSA Class I and Class II Kits (Immucor) when considered positive for either HLA class I 54 or II. Serum samples from patients with HLA-antibodies at week 6 were directly tested in SBA at 55 week 24. In both assays, antibodies bound to the beads were detected using a PE-conjugated goat-56 anti-human IgG detection antibody. In the SBA assay, all serum samples were EDTA treated prior to 57 luminex analysis. Samples were acquired on a luminex flow analyzer (LABScan 200) and data were 58 analyzed using the Match IT! Antibody software (version 1.3.1, Lifecodes, Immucor). LMX and SBA 59 results were assigned positive or negative according to the software assignment and the lot-specific 60 cut-off (Match IT! Antibody software). 61

CDC assay 62
A complement-dependent cytotoxicity (CDC) assay was performed on 4 patients (#4, #6, #9, and 63 #11), by incubation of patients' serum from baseline and week 6 with a cell panel encompassing all 64 mismatched HLA antigens. After incubation, propidium-iodide was added, in order to stain DNA of 65 lysed lymphocytes and measured with Patimed (Leica). 66

Fluorescence in situ hybridization 67
Fluorescence in situ hybridization (FISH) based on X-and Y-chromosome specific probes was 68 performed on formalin-fixed paraffin-embedded biopsies collected 6 weeks after MSC injection, to 69 investigate whether MSCs locally persisted in the tissue mucosa. Since this technique relies on 70 gender differences between patient and MSC-donor, 10/13 biopsies were analysed. One sample was 71 excluded due to technical problems. 72 H 2 O. Then, slides were treated with RNase (100 µg/mL) in PBS for 30 minutes at 37 °C and two times 77 washed for 3 minutes with PBS. After that, a short wash with H 2 O. Slides are then immersed for 15 78 minutes in 0.1% pepsin in 0.01 M HCl at 37 °C, followed by three times a 3-minute wash with PBS. 79 Slides were dehydrated in 70%, 90%, and 100% ethanol, respectively. Slides were air dried. 80 Thereafter, 10 µl probe mix (probe + hybridisation mix) were added to the slides. The probes 81 consisted of a biotin-labeled Y chromosome alpha satellite probe and a digoxigenin-labeled X 82 chromosome alpha satellite probe. The probes were denatured for 8 minutes on a 80 °C hot plate. 83 Slides were hybridized at 37 °C overnight in a humidity chamber with 50% formamide/2xSSC (pH = 84 7). 85

86
Post-hybridization protocol 87 The first post-hybridization wash twice 5 minutes with 2xSSC/0.1% Tween20 at 37 °C, followed by 88 twice a 5-minute wash with 50% formamide/2xSSC (pH = 7) at 44 °C and once a 5-minute wash 89 0.1xSSC at 60 °C. Then a short wash with MiliQ and a 3-minute wash with TBST. Incubated for 30 90 minutes with 1 µl streptavidin Cy-3 and mouse anti-dig FITC in 500 µl TNB at 37 °C. Followed by 91 three times a 3-minute wash with TBST and a short wash with PBS. Slides were dehydrated in 92 respectively, 70%, 90%, and 100% ethanol and air dried. Finally, slides were embedded in 93 DAPI/Citifluor (concentration: 500 ng/mL), 20 µl per slide, coverslip 24x60 mm. 94 95 Cytokine analysis 96 Colon biopsies from baseline and week 6 were analysed by Olink Proteomics (Uppsala, Sweden) 97 using the Explore 384 Inflammation panel. Proteins were extracted from the biopsies by 98 homogenizing in RIPA lysis buffer using 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1% 99 sodium deoxycholate, 1% Triton X-100 and 1x Complete Protease Inhibitors (Roche # 04693116001) 100 using Tissuelyser LT (Qiagen). Total protein content was determined using a BCA protein assay protein expression (NPX) values, using both internal and external controls for normalization, were 103 compared between baseline and week 6. Five samples were run in duplicates and showed 104 comparable results. 15% (54/369) of the analyzed proteins were not detected in our samples. One 105 sample did not pass quality control (#9, t=0) and was removed from further analysis, together with 106 the paired sample (#9, t=6). This sample also showed lower total protein concentrations compared 107 to the other samples. 108

Mass cytometry antibody staining and acquisition 122
The validated mass cytometry antibody panel is listed in Supplemental Table 10(2). Pre-conjugated 123 metal-labeled antibodies were purchased from Fluidigm (San Francisco, Calfifornia, USA) . The rest of 124 the antibodies were conjugated in-house using the MaxPar X8 antibody labeling kit (Fliudigm minutes with Fc Receptor blocking solution (Biolegend). Next, cells were stained with metal-129 conjugated antibodies (as listed in Supplemental Table 10) for 45 min at room temperature. After 130 staining, cells were washed three times with SB and incubated with 1 mL 500 mM iridium DNA 131 intercalator (Fluidigm) diluted in MaxPar Fix and Perm Buffer (Fluidigm) (1:4000) at 4°C overnight, or 132 up to 48 hours, to discriminate single-cells. Cells were washed three times with SB. Before data 133 acquisition, cells were diluted in distilled water containing 1:10 diluted EQ Four Element Calibration 134 Beads (Fluidigm Sciences). Samples were acquired at the Helios mass cytometer (Fluidigm, San 135 Francisco, CA, USA) with the narrow-bore injector at the Flow Core Facility (FCF) of the Leiden 136 University Medical Center. CyTOF data were acquired and analyzed on-the-fly, using dual-count 137 mode and noise-reduction on. All other settings were either default settings or optimized with 138 tuning solution, as instructed by Fluidigm. After data acquisition, the mass bead signal was used to 139 normalize the short-term signal fluctuations with the reference EQ passport P13H2302 during the 140 course of each experiment. Samples from the same patient were processed and acquired on the 141 same day. 142

143
Mass cytometry data analysis 144 First, normalized FCS files were checked for quality control of the staining and pre-gated for single 145 live CD45 + cells, removing duplicates, beads, dead cells, and debris in the FlowJo software V10.8.1. A 146 reference PBMC sample was included in each experiment to account for technical variation. All 147 reference PBMCs were obtained from blood from the same healthy individual. ComBat was applied 148 to align the PBMC reference samples and corresponding patient samples to correct for batch 149 effects(4). The markers CD66b, CD15, Nkp44, Nkp46, c-kit, CD40, CD80, and PD-L1 were not present 150 in the reference PBMCs at sufficient levels to scale and were thus not normalized. For further 151 analysis, these antibodies were not used for clustering but were shown for marker expression Cytosplore(5). First, major immune lineages ( Figure 4A) were identified at the overview level of a 5-155 level hierarchical stochastic neighbor embedding (HSNE) analysis on CD45 + cells from control and 156 case rectal biopsies, before and after 6 weeks of MSC therapy (3.4 million cells) with default 157 perplexity and iterations (30 and 1000, respectively)(6, 7). This global overview contained clusters of 158 myeloid cells (CD66b + granulocytes and CD66bmononuclear phagocytes), CD4 + T cells, CD8 + T cells, 159 double-negative (DN) T cells, γδ T cells, innate lymphoid cells (ILCs), B cells, and CD45 + Lineagecells. 160 CD66b + granulocytes and CD66bmyeloid cells were identified within the myeloid cell compartment 161 at the overview level of a 3-level HSNE analysis on 1.8 million cells. CD66bmyeloid cells were 162 further analyzed using tSNE. The other major immune lineage clusters were analyzed separately in a 163 data-driven manner up to a maximum number of 0.5×10 6 landmarks using t-distributed stochastic 164 neighbor embedding (tSNE)(7) with a default perplexity (30) Tables and Figures   Supplemental Table 1

Baseline
Week 2 Week 6 Week 24  The full Mayo score includes the following subscores: rectal bleeding score, stool frequency, physician's global assessment, and endoscopic mayo score. The clinical Mayo score is the full Mayo score without the endoscopic Mayo score.  Full Mayo score includes stool frequency, rectal bleeding, endoscopic findings, and physician's global assessment. The indicated thiopurines include azathioprine, 6-mercaptopurine, and tioguanine. a Patient 4, 7 and 9 used oral corticosteroids (budesonide 9 mg) at baseline. Patient 4 tapered with a dose of 9 mg every other day from week 8, patient 7 continued with a dose of 9 mg up to week 10 and patient 9 tapered with a dose of 9 mg every other day from week 6. R: rectal; O: oral; 5-ASA = 5-amino salicylic acid; CS: corticosteroids; TIO: thiopurines; MTX = methotrexate; ADA = adalimumab; VEDO: vedolizumab; IFX: infliximab; TAC = tacrolimus; TOFA: tofacitinib; USTE = ustekinumab; GOL: golimumab. * This patient first used infliximab and switched after 6 months to ustekinumab.

Baseline Week 2
Week 6 Week 12 Week 24 Week 6 Week 12 Week 24 The continuous scale (ranging from 0-22) of the Geboes score was used and biopsies were evaluated at week 6 and 24 compared to baseline, according to three categories: histological remission (GS ≤6), histological response (GS 7-12), and no response ( GS >12).    Figure 3. High-dimensional analysis of control and case biopsies at baseline and after 6 weeks of local MSC therapy. Frequencies of the CD66bmononuclear phagocytes, CD4 + T cells, double-negative T cells and, B cells from individual samples (control (n=12) and case biopsies (n=13) at baseline and case biopsies (n=12) at week 6); Each dot represents an individual sample. Red colored dot indicated mild inflammation (endoscopic Mayo score 1). ns, not significant. Wilcoxon signed-rank test was performed.