Recombinant protein expression and purification
NanoLuc is the tradename of an engineered deep-sea shrimp luciferase variant developed by Promega Corporation36. Genes encoding fusion proteins were inserted into pET-21d(+) vectors between NcoI and XhoI sites. IDO-Gal3 genetic and protein sequences are provided in Supplementary Information. Plasmids confer ampicillin resistance and were first transformed into One Shot TOP10 chemically competent E. coli (ThermoFisher), selected on Luria-Bertani (LB) ampicillin (50 μg ml−1) agar and incubated overnight at 37 °C. Isolated colonies from the plates were subcultured in 5 ml LB ampicillin broth overnight at 37 °C with orbital shaking. Successful transformants were confirmed by Sanger sequencing (Genewiz). Positive DNA sequences were then transformed into the kanamycin B resistant expression strain, Origami B (DE3) E. coli (Novagen) and selected on LB ampicillin with kanamycin B (15 μg ml−1) agar. Positive clones were picked, banked in LB with 10% w/v glycerol and used to preculture 5 ml of LB ampicillin and kanamycin. Overnight precultures were used to inoculate 1l 2× TY media (16 g tryptone, 10 g yeast extract, 5 g NaCl) with ampicillin and kanamycin B at 37 °C and 225 r.p.m. in an orbital shaker until approximate exponential growth phase (optical density (OD)600 nm = 0.6–0.8). IDO-Gal3 expression cultures were supplemented with 500 μM δ-Aminolevulinic acid (Sigma) at the time of inoculation. Recombinant protein expression was triggered using 0.5 mM isopropyl β–d-1-thiogalactopyranoside (ThermoFisher) and incubated for 18 h in an orbital shaker at 18 °C. Bacteria were washed and pelleted with PBS via centrifugation (13,180 × g at 4 °C for 10 min) with a superspeed centrifuge (ThermoFisher). Afterwards, the pellet was weighed, resuspended in 4 ml PBS per gram pellet with protease inhibitor (ThermoFisher) and disrupted by sonic dismembration (15 s on, 45 s off, 10 min cycling time; ThermoFisher). Following dismembration, lysates were treated with 20 units per gram pellet DNAse I (ThermoFisher) and 800 μg g−1 pellet lysozyme (ThermoFisher). The lysate was centrifuged (38,360 × g at 4 °C for 15 min) to remove the insoluble fraction and the supernatant was collected by decanting. Supernatant containing soluble recombinant protein was loaded into an α-lactose agarose (Sigma) affinity column pre-equilibrated with PBS. Columns were washed with 20 column volumes PBS and recombinant proteins were eluted with 100 mM β-d-lactose (Sigma) prepared in PBS. A final polishing step was performed by 200 kDa size exclusion chromatography on AKTA pure chromatography system (GE Life Sciences) to remove α-lactose and further isolate IDO-Gal3. Protein purity was determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE) and Coomassie staining. Endotoxin contaminants were removed by endotoxin removal solution (Sigma) following manufacturer instructions. Endotoxin content was analysed using Chromo-LAL kinetic chromogenic endotoxin quantification assay (Associates of Cape Cod) and determined to be below 0.1 EU ml−1 in all stocks.
IDO enzymatic activity assay
Recombinant human IDO expressed in E. coli was purchased from R&D Systems (predicted molecular mass of 46 kDa, a specific activity of >500 pmol min−1 μg−1 (or for equimolar calculations >29 pmol NFK min−1 pmol−1 IDO) as measured by its ability to oxidize l-tryptophan to N-formyl-kynurenine (NFK)). The specific activity of both proteins, IDO and IDO-Gal3, was measured before experiments to ensure maximal effect at the beginning of the assay, following the IDO manufacturer’s protocol. IDO-Gal3 was reacted in equimolar amounts to IDO in the standard protocol, and activities were compared using the unit pmol NFK min−1 pmol−1 IDO to more accurately compare activity.
IDO-Gal3 binding affinity
Affinity of IDO-Gal3 for lactose was determined using affinity chromatography in an AKTA Pure chromatography system (GE Life Sciences) equipped with consumer-packable glass column (GE Life Sciences) packed with α-lactose agarose affinity resin (Sigma-Aldrich). Proteins were eluted with a linear gradient of β-lactose (Sigma-Aldrich) in phosphate buffer.
IDO-Gal3 DNA sequence
The sequence used was CCATGGCGCACGCGATGGAAAACAGCTGGACCATCAGCAAAGAGTACCACATTGACGAGGAAGTTGGTTTCGCGCTGCCGAACCCGCAGGAAAACCTGCCGGACTTCTATAACGATTGGATGTTTATCGCGAAGCACCTGCCGGATCTGATTGAGAGCGGCCAGCTGCGTGAGCGTGTGGAAAAACTGAACATGCTGAGCATCGACCACCTGACCGATCACAAGAGCCAACGTCTGGCGCGTCTGGTTCTGGGTTGCATTACGATGGCGTACGTGTGGGGCAAAGGTCACGGCGACGTGCGTAAGGTTCTGCCGCGTAACATCGCGGTTCCGTACTGCCAACTGAGCAAGAAACTGGAACTGCCGCCGATTCTGGTGTATGCGGACTGCGTTCTGGCGAACTGGAAGAAGAAGGACCCGAACAAACCGCTGACCTATGAGAACATGGATGTGCTGTTCAGCTTTCGTGACGGTGATTGCAGCAAGGGCTTCTTTCTGGTGAGCCTGCTGGTTGAAATCGCGGCGGCGAGCGCGATCAAAGTGATTCCGACCGTTTTCAAGGCGATGCAGATGCAAGAGCGTGACACCCTGCTGAAAGCGCTGCTGGAAATCGCGAGCTGCCTGGAGAAGGCGCTGCAGGTGTTTCACCAAATTCACGATCACGTTAACCCGAAAGCGTTCTTTAGCGTGCTGCGTATCTACCTGAGCGGTTGGAAGGGCAACCCGCAGCTGAGCGACGGTCTGGTTTATGAGGGCTTCTGGGAAGATCCGAAAGAGTTTGCGGGTGGCAGCGCGGGTCAGAGCAGCGTGTTCCAATGCTTTGACGTTCTGCTGGGCATTCAGCAAACCGCGGGTGGCGGTCATGCGGCGCAGTTCCTGCAAGATATGCGTCGTTACATGCCGCCAGCGCACCGTAACTTCCTGTGCAGCCTGGAAAGCAACCCGAGCGTGCGTGAGTTTGTTCTGAGCAAAGGTGACGCGGGCCTGCGTGAAGCGTATGATGCGTGCGTGAAGGCGCTGGTTAGCCTGCGTAGCTACCACCTGCAGATCGTTACCAAATATATCCTGATTCCGGCGAGCCAGCAACCGAAAGAAAACAAGACCAGCGAGGACCCGAGCAAACTGGAGGCGAAGGGTACCGGCGGTACCGATCTGATGAACTTTCTGAAGACCGTGCGTAGCACCACCGAGAAGAGCCTGCTGAAAGAGGGTGGATCCGGCGGCGGCAGCGGCGGCAGCGGCGGCAGCGGCGGCGAATTCGCGGACAACTTCAGCCTGCACGATGCGCTGAGCGGTAGCGGTAACCCGAACCCGCAGGGTTGGCCGGGTGCGTGGGGTAACCAACCGGCGGGTGCGGGTGGCTACCCGGGTGCGAGCTATCCGGGTGCGTATCCGGGTCAGGCTCCGCCGGGTGCGTACCCGGGCCAAGCTCCGCCGGGTGCTTATCCTGGTGCGCCGGGCGCGTACCCGGGTGCGCCGGCGCCGGGCGTGTACCCGGGTCCGCCGAGCGGTCCGGGCGCGTATCCGAGCAGCGGCCAGCCGAGCGCGCCGGGTGCGTATCCGGCGACCGGCCCGTATGGTGCGCCGGCGGGTCCGCTGATTGTTCCGTATAACCTGCCGCTGCCGGGTGGCGTGGTTCCGCGTATGCTGATCACCATTCTGGGCACCGTGAAGCCGAACGCGAACCGTATCGCGCTGGACTTCCAACGTGGTAACGATGTTGCGTTCCACTTTAACCCGCGTTTTAACGAGAACAACCGTCGTGTGATTGTTTGCAACACCAAACTGGACAACAACTGGGGCCGTGAGGAACGTCAGAGCGTGTTCCCGTTTGAGAGCGGCAAGCCGTTCAAAATTCAAGTGCTGGTTGAACCGGACCACTTTAAGGTGGCGGTTAACGATGCGCACCTGCTGCAGTACAACCACCGTGTTAAGAAACTGAACGAAATCAGCAAACTGGGCATCAGCGGTGACATTGATCTGACCAGCGCGAGCTATAACATGATTCTCGAG, with restriction sites underlined.
Amino acid sequence
MAHAMENSWTISKEYHIDEEVGFALPNPQENLPDFYNDWMFIAKHLPDLIESGQLRERVEKLNMLSIDHLTDHKSQRLARLVLGCITMAYVWGKGHGDVRKVLPRNIAVPYCQLSKKLELPPILVYADCVLANWKKKDPNKPLTYENMDVLFSFRDGDCSKGFFLVSLLVEIAAASAIKVIPTVFKAMQMQERDTLLKALLEIASCLEKALQVFHQIHDHVNPKAFFSVLRIYLSGWKGNPQLSDGLVYEGFWEDPKEFAGGSAGQSSVFQCFDVLLGIQQTAGGGHAAQFLQDMRRYMPPAHRNFLCSLESNPSVREFVLSKGDAGLREAYDACVKALVSLRSYHLQIVTKYILIPASQQPKENKTSEDPSKLEAKGTGGTDLMNFLKTVRSTTEKSLLKEGGSGGGSGGSGGSGGEFADNFSLHDALSGSGNPNPQGWPGAWGNQPAGAGGYPGASYPGAYPGQAPPGAYPGQAPPGAYPGAPGAYPGAPAPGVYPGPPSGPGAYPSSGQPSAPGAYPATGPYGAPAGPLIVPYNLPLPGGVVPRMLITILGTVKPNANRIALDFQRGNDVAFHFNPRFNENNRRVIVCNTKLDNNWGREERQSVFPFESGKPFKIQVLVEPDHFKVAVNDAHLLQYNHRVKKLNEISKLGISGDIDLTSASYNMILEHHHHHH
LPS administration at the hock as a model of inflammation
B6 mice were administered 2.1 μg IDO-Gal3 in 40 μl PBS subcutaneously (ipsilateral and contralateral) at the region of the hock and challenged with 2 ng g−1 lipopolysaccharide (LPS) in 40 μl, after 24 h or 120 h post IDO-Gal3 treatment, to assess local and distal modulation of inflammation. At 2 h after LPS administration, animals were euthanized and the injection site collected.
Quantitative PCR
Soft tissue was separated from bone and stored in RNAlater RNA stabilization reagent (Qiagen) in preparation for qPCR. Soft tissues were homogenized and RNA purified using RNeasy Protect mini kit (Qiagen). Complementary DNA was synthesized from RNA using the High-Capacity cDNA Reverse Transcriptase kit (ThermoFisher) for use in qPCR in accordance with manufacturer instructions. qPCR analysis was run with primers specific for pro-inflammatory cytokines (IL-12a, IL-12b, IL-1β, IFN-γ and IL-6) using Applied Biosystems QuantStudio 12K Flex Real-Time PCR System. Results are presented as the ratio of gene expression to glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) expression determined by the relative quantification method. Treatment groups were normalized to PBS only group.
Housing conditions
Mice were housed in individually ventilated cages with corn cob bedding, provided reverse osmosis water and standard diet, 14 h:10 h dark:light cycle, 68–79 ℉ room temperature and 30–70% relative humidity.
Hock infiltration
Tissues were fixed with 10% neutral formalin (pH 7.4) overnight. After fixation, tissues were washed in deionized water and decalcified by storing in 10% ethylenediaminetetraacetic acid (EDTA) at 4 °C for 3 weeks. Samples were assessed every 2–3 d for stiffness and EDTA solution replenished. Tissues were submitted in 70% ethanol to the University of Florida Molecular Pathology Core for processing, paraffin embedding, sectioning, mounting and staining with haematoxylin and eosin (H&E). Tissues were imaged using a Zeiss Axiovert 200M with a ×20 objective lens through the multidimensional acquisition module. Eleven images were taken per tissue and scored by two blinded independent individuals on the basis of cellular infiltration (0: absent, 1: mild, 2: moderate, 3: severe) and epidermis hypertrophy (0: 0–20 μm in thickness, 1: 21–40 μm, 2: 41–60 μm, 3: 61 μm or above).
Mass spectrometry
IDO-Gal3 (2.2 μg) in 40 μl PBS or 40 μl PBS alone was injected into the hock site of B6 mice (n = 3). At 30 min after injection, the mice were euthanized, the hock and tibia tissue regions excised and flash frozen using liquid nitrogen. These samples were then submitted to the Southeast Center for Integrated Metabolomics at the University of Florida for mass spectrometric analysis of kynurenine levels.
Serum cytokine analysis
At the prescribed endpoints, animals were placed under deep terminal anaesthesia and cardiac punctures performed. Blood (2 ml) was collected in microtainer serum separator tubes and centrifuged at 1,500 r.p.m. for 5 min at room temperature. The serum was collected and stored at −20 °C. Cytokine analysis was performed using a Luminex 200 system running xPONENT 3.1 software (Luminex) following manufacturer instructions.
Imiquimod-induced psoriasis
Pre-clinical modelling of psoriasis was carried out in 8–12-week-old female C57BL/6j mice (Jackson Laboratory) in accordance with the Institutional Animal Care and Use Committee (IACUC) at the University of Florida. The model used was a modified version of a previously reported model37. Briefly, mice were anaesthetized and their backs were shaved, followed by application of depilatory cream to remove any remaining fur. Each day for 14 d total, 5% IMQ cream (62.5 mg, Patterson Veterinary Supply, 07-893-7787) was applied to the backs of the mice. On the 3rd day of IMQ application, mice were subcutaneously injected with five 10 μg doses of IDO-Gal3 in sterile saline distributed evenly throughout the back, a molar equivalent of NL-Gal3 delivered in the same volume, or a sterile saline control (n = 12 per group). Disease severity was measured each day using a modified version of the Psoriasis Area and Severity Index (PASI) where area of effect is not taken into account. Erythema (redness), scaling and thickening were scored independently and assigned a score on a scale of 0 to 4: 0, none, 1: slight, 2: moderate, 3: marked, 4: very marked. The cumulative score was reported as a measure of the severity of inflammation (scale 0–12).
Near-infrared conjugation and in vivo imaging of IDO-Gal3 injected subcutaneously
In vivo imaging of fluorescently tagged IDO-Gal3 was carried out in 8–12-week-old female C57BL/6j mice (Jackson Laboratory) in accordance with the IACUC at the University of Florida. Before injection, IDO-Gal3 was incubated with IRDye 680RD NHS Ester (LI-COR, 929-70050) following manufacturer instructions. Mice received a 40 μl (2.1 μg, 3.27 μM) subcutaneous injection of fluorescently labelled IDO-Gal3 to the hock. Immediately following injection and every subsequent 24 h following injection, mice were imaged using an IVIS Spectrum in vivo imaging system (PerkinElmer). Fluorescent images were captured using the AF680 emission filter, subject size 1.5 cm, 0.2 s exposure time, field of view B (6.6 cm), medium binning (factor of 8) resolution and a 1F/Stop aperture. Relative fluorescent intensities were represented by a pseudo colour scale ranging from red (least intense) to yellow (most intense).
In vivo imaging of tissue distribution after hock injection
NanoLuc-Gal3 (164 pmol) in 40 μl PBS was injected subcutaneously into the hock of B6 mice. At the prescribed time points, animals were euthanized in accordance with approved protocols. Organs and tissues of interest were collected, weighed, processed, incubated with furimazine and bioluminescence quantified by a luminometer. Bioluminescence images were acquired using an IVIS Spectrum in vivo imaging system. Living Image software v4.3.1 (PerkinElmer) was used to acquire the data immediately after furimazine administration. Exposure time for the bioluminescence imaging was 1 s. Regions of interest (ROIs) were quantified as the average radiance (photons s−1 cm−2 sr−1). The specific amount of protein in tissue was determined by comparison with a standard curve of NanoLuc-Gal3 activity.
Oral Listeria infection after hock injection
Immune suppression from IDO-Gal3 was evaluated in response to oral infection with Listeria monocytogenes InIAM (strain 10403s). Tissue bacteria burdens in the liver and spleen were used as a metric for infection where the liver is local for gut infection and the spleen represents systemic spread of infection. On the day before infection, 10-week-old naïve C57BL/6 mice (Taconic Biosciences) received IDO-Gal3 injection subcutaneously (hock), while control mice received sterile saline injection. The following day, mice were orally infected with 2 × 109 colony-forming units (c.f.u.) of Listeria monocytogenes per mouse. To prepare for infection, bacteria were cultured overnight in BHI broth at 37 °C, shaking at 220 r.p.m. Before infection, a subculture of 2 ml of the bacteria and 18 ml BHI broth was cultured under the same conditions until reaching an OD600 = 0.8. Bacteria were pelleted, resuspended in 500 μl of sterile PBS, and 50 μl of bacteria was pipetted onto a small square of white bread and fed to each mouse individually. Once the entire piece of bread was eaten, mice were returned to their cage. At day 7 post infection, mice were euthanized, followed by collection of the spleen and liver. Tissue was homogenized, suspended in 1% saponin for 1 h, then plated at serial dilutions from undiluted to 1:1,000. BHI Agar plates were treated with streptomycin to limit non-specific bacterial growth, as this strain of L. monocytogenes is streptomycin resistant. C.f.u.s were counted at 24 h post-plating.
Murine model of periodontal disease
All mice were lavaged with 25 μl 0.12% chlorhexidine gluconate (3 M) for 3 d. On days 4, 5, 6 and 7, mice received a 25 μl oral lavage with 2.5 × 109 P. gingivalis strain 381 and 2.5 × 109 Aggregatibacter actinomycetemocomitans strain 29522 (ATCC) resuspended in 2% low-viscosity carboxy-methyl-cellulose (Sigma-Aldrich). Oral lavage was repeated every week for 5 weeks. Each week, 1 d before the first day of infection (prophylactic) or 1 d after the last day of infection (therapeutic), 10 μl of IDO-Gal3 was injected into the submandibular space using a 30-gauge insulin syringe (Becton Dickenson). Each week of infection, on the first day of infection, before infection, microbial sampling of the oral environment was performed with calcium alginate swabs (ThermoFisher). One week following the last infection, the mandibles were collected to evaluate soft tissue soluble mediator expression and bone morphometric analysis.
Oral soft tissue cytokine expression
Mandibles with both soft tissue and bone were subjected to bead beating at two 2 min intervals with 2 min of cooling in between using 1.0-mm-diameter zirconia silica beads (BioSpec) in cell extraction buffer (ThermoFisher). The buffer was prepared with a protease inhibitor cocktail (mini cOmplete, Roche) and PMSF protease inhibitor (Abcam) to allow for dissociation and lysis of all soft tissue while leaving the hard tissues intact. MILLIPLEX Multiplex assays (EMD Millipore) were used to probe resulting lysates for IL-6, IL-1β, IL-10 and MCP1 according to manufacturer protocols. Data were acquired on a Luminex 200 system running xPONENT 3.1 software (Luminex) and analysed using a 5-paramater logistic spline-curve fitting method using Milliplex Analyst v5.1 software (Vigene Tech). Data are presented as pg ml−1 normalized to total protein (BCA assay; ThermoFisher Pierce).
Mandible bone morphometric analysis
Mandibles were fixed in 4% buffered formalin for 24 h, stored in 70% alcohol and scanned at 18 μm resolution using a micro-CT system (Skyscan). Three-dimensional images were reconstructed and the resulting images re-oriented spatially using anatomical landmarks with the NRecon and DataViewer software (Skyscan). A standardized 5.4 mm3 ROI was set with standardized dimensions of 1.5 (frontal) × 4.0 (sagittal) × 0.9 mm (transversal). Anatomical landmarks were used for the standardized positioning of the ROI: frontal plane, the roof of the furcation area between mesial and distal roots of the upper first molar; sagittal plane, anterior limit was the distal aspect of the mesial root of the first molar. The thickness of the ROI on the transversal plane was set to 50 slices (900 μm) and counted towards the palatal /medial direction beginning from the image that included the centre of the upper first molar in its transversal width. A standardized threshold was set to distinguish between non-mineralized and mineralized tissues where total volume and total thickness of the ROI were calculated. Mean trabecular bone volume (BV) and thickness (BT) analysis assessed the percentage of mineralized tissue (BV; BT) within the total volume/thickness (TV; TT) of the ROI and is presented as the BV/TV ratio (mean trabecular bone volume) or BT/TT ratio (mean trabecular thickness). Vertical bone loss is the distance from the cementoenamel junction (CEJ) to the alveolar bone and was calculated at 12 sites over three molars and averaged to calculate the average vertical bone loss in mm.
Oral bacterial burden
Genomic DNA was isolated from microbial sampling of the oral environment using a DNeasy kit (Qiagen) according to manufacturer instructions. The gDNA was then probed for P. gingivalis 16S, A. actinomycetemocomitans 16S and total 16S using real-time PCR. The percentage of A. actinomycetemocomitans 16S and P. gingivalis 16S within the total 16S compartment was calculated using the following formula: Ct value of total 16S/Ct value of A. actinomycetemocomitans or P. gingivalis 16S. 16s rRNA For: AGA GTT TGA TCC TGG CTC AG; Rev: ACG GCT ACC TTG TTA CGA CTT; Pg For: CTT GAC TTC AGT GGC GGC AG; Rev: AGG GAA GAC GGT TTT CAC CA; Aa For: GTT TAG CCC TGG CCG AAG; Rev: TGA CGG GCG GTG TGT ACA AGG.
Mechanical overloading osteoarthritis murine model
IDO-Gal3 activity was assessed in a post-traumatic osteoarthritis mouse model30 that applies cyclic mechanical loading to the knees of aged (6 months) mice, causing mechanical damage and consequent inflammation and cartilage degradation30,31. The mice were anaesthetized and placed in a fixture with the knee in flexion; loading (9 N) was applied axially for 500 cycles and loading sessions were done on the mice 5 times per week during the experiment. IDO or IDO-Gal3 was prepared at 143 μM, and 20 μl of each treatment was injected intra-articularly at the start of each week of the study, with each knee receiving 4 total treatments. At the end of the 4-week study, mice were euthanized for analysis of gene expression and histopathology.
Mechanical overloading osteoarthritis pharmacokinetics analysis
Pharmacokinetics of IDO and IDO-Gal3 retention after local injection at the disease site was assessed by intravital imaging over the course of 7 d. Both proteins were labelled with Li-Cor IRDye 680RD NHS ester (Li-Cor Biosciences) to visualize and measure protein knee retention. Mice were subjected to mechanical loading for 2 weeks before each treatment was administered via intra-articular injection. Intravital imaging was performed (excitation 672 nm, emission 694 nm) immediately following injection and every subsequent 24 h. The fluorescence signal measured at the joint over time was normalized to the initial measurement for each knee and an exponential decay was individually fit for each specimen. The AUC/bioavailability for each joint was calculated from the best-fit line of exponential decay. After 7 d, the mice were euthanized and an ex vivo image was taken of each joint with the surrounding skin removed to increase measurement sensitivity.
Mechanical overloading osteoarthritis, gene expression analysis
Gene expression was evaluated by TaqMan qPCR in the knee joint and the popliteal lymph node that drains the knee joint. Following euthanization, knees and popliteal lymph nodes were excised. Combined joint tissue from the synovial wall and articular surface (not exceeding 30 mg total) and the popliteal lymph nodes were homogenized with bead pulverization in Qiazol. RNA was extracted and purified using the RNeasy Plus mini kit from Qiagen and quantified using NanoQuant plate from Tecan in a microplate reader (Tecan Infinite 500, Tecan). The RNA was converted to cDNA using the iScript Synthesis kit from Bio-Rad (Hercules). Gene expression was calculated by the ΔΔCt method, normalizing to GAPDH and beta-actin (ACTB). TaqMan reagents were purchased from ThermoFisher and used according to provided protocols, using appropriate primers (IL-12β: Mm01288989_m1, IL-6: Mm01210732_g1, MMP13: Mm00439491_m1, TNF-α: Mm00443258_m1, GAPDH: Mm99999915_g1, ACTB: Mm02619580_g1).
Mechanical overloading osteoarthritis, histologic staining and scoring
Tissue samples were fixed in 10% formalin and decalcified in 20% EDTA for 7 d. A standard 8 h cycle of graded alcohols, xylenes and paraffin wax was used to process tissues before embedding and sectioning at 5 μm thickness. Sections were mounted on positively charged glass slides and stained with H&E using the Gemini autostainer (ThermoFisher). Safranin O staining was performed using the StatLab staining kit. Sections were imaged using Leica SCN400 slide scanner. Each joint was evaluated by at least two mid-frontal sections for both H&E and safranin O stains. A board-certified veterinary pathologist conducted histopathologic interpretations under blinded conditions38. OARSI scoring was based on medial and lateral tibial plateaus (scale of 0–6)39, and a generic score was concurrently assigned on the basis of H&E features and the safranin O staining of the tibial plateaus according to the DJD methodology (DJD severity, scale 0–3).
Surgically induced osteoarthritis, rat model
Sixteen male Lewis rats (12–14-week-old, approximately 250 g) were acquired from Charles River Laboratories. Rats were acclimated to the University of Florida housing facilities for 1 week. After acclimation, rats underwent baseline gait and von Frey behavioural testing. Following baseline behavioural testing, all rats received medial collateral ligament plus medial meniscus transection (MCLT + MMT) surgery to their right hind limb. Gait data were collected on weeks 3, 5 and 7 post-surgery using a Phantom Miro Lab320 (Phantom Camera Control 3.0), while von Frey testing was conducted weekly after surgery. At 8 weeks post-surgery, rats received a unilateral saline or IDO-Gal3 injection. Rats underwent von Frey testing the day following injection and gait testing 2 d after injection. Both gait and von Frey data were collected weekly until euthanasia.
Surgically induced osteoarthritis, behavioural study
MCLT + MMT surgery rats were anaesthetized in a 2.5% isoflurane (Patterson Veterinary) sleep box. Rats were then aseptically prepped with betadine surgical scrub (Purdue Products) and 70% ethanol in triplicate, and transferred to a sterile field with anaesthesia maintained via mask inhalation of 2.5% isoflurane. During MCLT + MMT surgery, a 1–2 cm midline skin incision was made along the medial aspects of the rat knee and the skin was retracted to reveal the medial collateral ligament. The medial collateral ligament was transected and the knee was placed in the valgus orientation to stretch the medial compartment and expose the medial meniscus. The medial meniscus was cut radially, and absorbable 5–0 vicryl braided sutures (Ethicon) were used for muscle closure and 4–0 ethilon nylon monofilament sutures (Ethicon) were used for skin closure. Rats recovered post-operatively in a warming box until weight bearing on all limbs. For pain management, rats received a subcutaneous injection of buprenorphine (0.05 mg kg−1) (Patterson Veterinary) intra-operatively and every 12 h for 48 h. Rats were grouped into a saline injection cohort (n = 7) or IDO-Gal3 injection cohort (n = 8). At 8 weeks after OA induction, rats received 30 µl unilateral injections of either sterile saline or IDO-Gal3 (33 pg µl−1) in the operated knee using sterile allergy syringes (1 ml, 27G × 3/8) (Becton Dickinson). First, rats were anaesthetized using 2.5% isoflurane (Patterson Veterinary) and the operated knee was aseptically prepped as described above. Then the needle was inserted through the patellar ligament following the patellar groove into the joint space. The knee was flexed and the injection site was cleaned with sterile gauze and 70% ethanol.
Surgically induced osteoarthritis, tactile sensitivity
Tactile sensitivity was assessed by measuring the 50% paw withdrawal threshold determined using the Chaplan up-down method for von Frey filaments1. Before surgery, animals were acclimated to the wire mesh-floored cage. Rats underwent von Frey testing the week before surgery and weekly after surgery for 12 weeks. On week 8, rats were injected, then underwent von Frey testing the following day. During von Frey testing, a von Frey filament series (0.6, 1.4, 2.0, 4.0, 6.0, 8.0, 15.0 and 26.0 g) was applied to the plantar region of each hind foot. First, the 4.0 g von Frey filament was applied. A less-stiff filament was applied if a paw withdrawal occurred, and a stiffer filament was applied if a paw withdrawal did not occur. Using these data, the force at which rats were equally likely to withdraw or tolerate was calculated via Chaplan’s approximation1.
Surgically induced osteoarthritis, magnetic capture
Commercially available streptavidin-functionalized particles (Dynabeads MyOne Streptavidin C1, 65001; Life Technologies) were coated with biotinylated antibodies for CTXII (AC-08F1, ImmunoDiagnostic Systems), IL-6 and MCP1 (517703 and 505908, Biolegend). Here, particles were washed 3 times in PBS, incubated for 2 h on a tube revolver at room temperature in antibody mixes that contained either 8.9 ng µl−1 anti-CTXII, 8.9 ng µl−1 anti-IL-6 or 357 ng µl−1 anti-MCP1 and then moved to static incubation at 4 °C overnight. Particles were then washed three times in PBS containing 2% BSA and 2 mM EDTA (capture buffer), with final antibody amounts per particle measured to be 0.33 ng Ab µg−1 particle (anti-CTXII), 0.33 ng Ab µg−1 particle (anti-IL-6) and 13.0 ng Ab µg−1 particle (anti-MCP1).
Following euthanasia via exsanguination, magnetic capture was performed in both the operated and contralateral knee, as previously described40. Briefly, 300 µg of antibody-conjugated magnetic particles (equal parts for each particle type) were suspended in 10 µl of saline, then injected in the operated and contralateral knee. After 2 h incubation in the knee, particles were collected via 5 repeated 50 μl PBS washes of the knee, with collected fluid combined and particles in the fluid isolated via a magnetic separator. Collected particles were washed twice with capture buffer, then incubated for 15 min in the 100 mM Glycine-Tris buffer (pH 3.1) containing 2% BSA and 2 mM EDTA (release buffer). Following biomarker release, magnetic particles were isolated by magnetic separation and the pH of the supernatant was adjusted to 8.3 for enzyme-linked immunosorbent assays (ELISA). CTXII was then measured in the supernatant using the Cartilaps CTXII ELISA kit (AC-08F1, ImmunoDiagnostic Systems Cartilaps kit) according to manufacturer instructions, and CCL2 was quantified using a rat CCL2 ELISA kit (KRC1012 Life Technologies) according to manufacturer instructions and previously described modifications41. IL-6 was quantified using ELISA developed in the laboratory using biotin anti-rat IL-6 antibody (517703) and purified (coating) anti-rat IL-6 antibody (517701, Biolegend). Coating antibody was diluted in 100 mM NaHCO3 and 34 mM Na2CO3 (pH 9.5), placed in coated microwells (Nunc MaxiSorp, 434797; ThermoFisher), incubated for 30 min on a plate shaker then overnight at 4 °C, washed 5 times with PBS with 0.5% Tween-20 (wash buffer), blocked for 1 h with PBS containing 2% BSA and 10% heat-treated bovine serum, and finally washed 5 times with wash buffer. Samples and standards were pre-incubated with anti-IL-6 detection antibody (30 min at room temperature and then overnight at 4 °C), then added to ELISA plate and incubated for 3 h. The plate was then washed 5 times, and 100 μl of avidin-HRP (405103, Biolegend, diluted 500 times in 2% BSA and 10% heat-treated bovine serum) was added to the plate and incubated for 30 min. The plate was again washed 5 times, with 100 μl of tetramethylbenzidine substrate added for 15 min, followed by 100 μl of stop solution (diluted sulfuric acid). Absorbance was read at 450 and 650 nm. Particles were quantified in 60 μl of capture buffer, with particle suspensions read for absorbance at 450 nm using Synergy 2 Multi-Mode microplate reader, as previously described40.
Surgically induced osteoarthritis histology
Following magnetic capture, operated and contralateral knees were dissected and placed in 10% neutral buffered formalin (ThermoFisher) for 48 h at room temperature. Following fixation, knees were decalcified using Cal-Ex (ThermoFisher) for 5 d at room temperature, dehydrated through an ethanol ladder and embedded in paraffin wax via vacuum infiltration. Then, 10 µm frontal sections were acquired, with at least one section taken at every 100 µm through the loading region of the medial meniscus. Slides were stained with toluidine blue.
Surgically induced osteoarthritis-joint-retention study
Sixteen male Lewis rats (12–14-week-old, approximately 250 g) were acquired from Charles River Laboratories. Rats were acclimated to the University of Florida housing facilities for 1 week. Then rats underwent MCLT + MMT surgery. After 8 weeks, rats received a 50 µl injection of either Nano-Glo Luciferase (NL) or Nano-Glo Luciferase Galectin-3 (NL-Gal3). Rats were IVIS imaged immediately following injection, then at 1, 2, 4, 8, 12, 16, 20, 24 and 28 d post-injection. After imaging on day 28, rats were euthanized, and both operated and contralateral knees were dissected for joint tissue distribution analysis via IVIS. In vivo joint retention was measured in the operated and contralateral knees of 16 male Lewis rats. At 8 weeks post MCLT + MMT surgery, knees were aseptically prepped and rats received bilateral injections of NL (50 µl, 3.27 µM) (n = 8) or bilateral injections of NL-Gal3 (50 µl, 3.27 µM) (n = 8) in both operated and contralateral knees. (Surgeries and injections were conducted the same way as described above). Knees were flexed, then injected with 50 µl furimazine (50 µl, 1:50 dilution in PBS). Immediately after injection, knees were flexed again and luminescence was measured with IVIS using a 1 s (and 60 s) exposure time in field of view D. Furimazine injections and IVIS imaging were repeated at 1, 2, 4, 8, 12, 16, 20, 24 and 28 d post NL or NL-Gal3 injection. For analysis, a ROI was drawn around the largest luminescent signal and copied to create an identical sized ROI for all knees.
After imaging on day 28, rats were euthanized via cardiac puncture under deep isoflurane anaesthesia. Knees were dissected to isolate the patellar tissue, tibial tissue, femoral tissue and meniscus. The patellar tissue included the patella, patellar synovium and fat pad. The tibial tissue included the tibia and attached synovium. The femoral tissue included the femur and attached synovium. Tissues were placed in 24-well plates and incubated with furimazine (1:50 dilution in PBS) at room temperate for roughly 1 min. After incubation, tissues were removed from the 24-well plates and luminescence was measured with IVIS using auto-exposure in field of view C. For assessment, individual ROIs were drawn around the patellar tissue, tibial tissue, femoral tissue and meniscus. The ROI for each tissue was copied to create an identical sized ROI for the respective tissue.
Statistical analysis
Statistical analyses were performed using GraphPad Prism 8 and 9, with the following exceptions. Surgically induced osteoarthritis data were analysed using R Analytics 4.0.4 with RStudio 2022, and MATLAB 2020b was used for Mann-Whitney U-tests on psoriasis data. Study-specific analyses are reported in figure captions.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.