Research ArticleDrug Development

A potent peptide-steroid conjugate accumulates in cartilage and reverses arthritis without evidence of systemic corticosteroid exposure

See allHide authors and affiliations

Science Translational Medicine  04 Mar 2020:
Vol. 12, Issue 533, eaay1041
DOI: 10.1126/scitranslmed.aay1041

Peptide targeting prevents toxicity

Treatments that target anti-arthritic drugs to joints can help avoid systemic toxicity. Cook Sangar et al. identified a cystine-dense peptide that accumulates in cartilage. Imaging agents conjugated to the peptide distributed to cartilage in rodents when delivered systemically and bound to human cartilage explants ex vivo. Conjugating the steroid triamcinolone acetonide to the peptide alleviated inflammation in the joints of a rat model of collagen-induced rheumatoid arthritis without systemic toxicity, suggesting that the conjugate could have potential joint-targeting therapeutic utility.


On-target, off-tissue toxicity limits the systemic use of drugs that would otherwise reduce symptoms or reverse the damage of arthritic diseases, leaving millions of patients in pain and with limited physical mobility. We identified cystine-dense peptides (CDPs) that rapidly accumulate in cartilage of the knees, ankles, hips, shoulders, and intervertebral discs after systemic administration. These CDPs could be used to concentrate arthritis drugs in joints. A cartilage-accumulating peptide, CDP-11R, reached peak concentration in cartilage within 30 min after administration and remained detectable for more than 4 days. Structural analysis of the peptides by crystallography revealed that the distribution of positive charge may be a distinguishing feature of joint-accumulating CDPs. In addition, quantitative whole-body autoradiography showed that the disulfide-bonded tertiary structure is critical for cartilage accumulation and retention. CDP-11R distributed to joints while carrying a fluorophore imaging agent or one of two different steroid payloads, dexamethasone (dex) and triamcinolone acetonide (TAA). Of the two payloads, the dex conjugate did not advance because the free drug released into circulation was sufficient to cause on-target toxicity. In contrast, the CDP-11R–TAA conjugate alleviated joint inflammation in the rat collagen–induced model of rheumatoid arthritis while avoiding toxicities that occurred with nontargeted steroid treatment at the same molar dose. This conjugate shows promise for clinical development and establishes proof of concept for multijoint targeting of disease-modifying therapeutic payloads.


More than 2 million Americans live with chronic pain and limitations in physical activity from diseases such as rheumatoid arthritis (RA) that cause multijoint inflammation (1). Current therapies are mechanistically capable of reducing inflammation and concomitant cartilage degradation, but to achieve these benefits, they also cause severe toxicities in other tissues and increase the risk of infection in patients (26). Maximizing delivery of therapeutic agents to joints while minimizing systemic exposure would improve the therapeutic window of agents that have proven efficacy, but that cannot be used for prolonged treatment due to on-target toxicity in healthy tissues.

We describe the discovery of a series of cystine-dense peptide (CDPs) that rapidly accumulate in cartilage after systemic administration in rodents. CDPs are a class of miniproteins that are defined by having at least three disulfide bonds (cystines) in a protein that is typically 20 to 60 amino acids in length (7). They are frequently found in the venom of spiders, snakes, and scorpions as biodefense mechanisms, but they are also made by a wide variety of other species, including plants, fungi, bacteria, marine mollusks, and insects (810). Because of their rigid three-dimensional structure and their enormous chemical diversity, CDPs are well suited to drug discovery applications. The exceptional stability of the cystine-knot fold makes some of these peptides highly resistant to thermal, chemical, and proteolytic degradation (7, 11, 12). We previously used a CDP derived from the venom of the death stalker scorpion to develop the clinical candidate tozuleristide, a modified chlorotoxin peptide conjugated to a near-infrared dye (13, 14). This CDP-dye conjugate accumulates in tumors after systemic administration and facilitates real-time fluorescence imaging of tumor tissue during surgery (15).

After identification and characterization of cartilage-accumulating CDPs, we found that a CDP successfully delivered chemically divergent payloads, including a fluorophore and corticosteroids, to cartilage. To establish the utility of the cartilage-accumulating CDP therapeutic approach, we conjugated a steroid to a CDP by a labile chemical linker and showed that systemic administration of the construct reversed joint inflammation in the collagen-induced rat model of RA while protecting sensitive tissues from toxicities that occurred with nontargeted steroid treatment at the same dose.


CDPs accumulate and persist in cartilage

On the basis of the development of the tumor-homing modified chlorotoxin CDP as a potential oncology surgical aid (13, 14), we hypothesized that other CDPs may have therapeutic utility stemming from their native properties. To identify target tissues, we performed an in vivo biodistribution screen of 42 CDPs from 20 species of origin using quantitative whole-body autoradiography (QWBA) of mice. Our screen used sequences selected from the UniProt database that were annotated as toxins and contained CDP regions. From the biodistribution survey, we made the marked observation that several CDPs with diverse amino acid sequences accumulated in cartilaginous regions throughout the body, including hyaline cartilage of the synovial joints and fibrocartilage of the intervertebral discs (IVDs), sites of arthritic conditions in humans (Fig. 1, A to D). Three hours after intravenous administration, three cartilage-targeting CDPs (14C-CDP-11R, 14C-CDP-09R, and 14C-CDP-45R) reached concentrations ranging from 3.9 ± 0.3 to 8.1 ± 1.7 nmol/g in the knee and 3.4 ± 0.1 to 5.9 ± 1.3 nmol/g in the IVD of mice (Fig. 1E). In contrast, most other CDPs screened show very low accumulation in cartilaginous tissues, exemplified by 14C-CDP-29 and 14C-CDP71R, for which signal ranged from undetected to 0.26 ± 0.12 nmol/g in the knee and 0.1 ± 0.09 to 0.29 ± 0.01 nmol/g in the IVDs at 3 hours (Fig. 1E). CDP-11R was selected for further pharmacokinetic and proof-of-concept studies based on its apparent high accumulation in cartilage.

Fig. 1 Cartilage accumulation of CDPs after systemic administration in mice.

(A to D) Representative autoradiographs of 14C-CDP-11R, 3 hours after intravenous administration. Cartilaginous tissue is highlighted with light blue (hip), dark blue (knee), green (IVD), red (shoulder), or magenta (elbow) arrowheads. Scale bars, 1 mm (A to C) and 5 mm (D). (E) Quantification of radioactive signal in the IVD or knee (n = 2), 3 hours after intravenous administration of 14C-CDP-71R (9.6 μCi, 77.2 nmol), 14C-CDP-09R (7.4 μCi, 59.5 nmol), 14C-CDP-45R (8.1 μCi, 65.1 nmol), 14C-CDP-29 (12 μCi, 97 nmol), or 14C-CDP-11R (7.5 μCi, 60 nmol). (F and G) Quantification of 14C-CDP-11R signal over 96 hours in knee and IVD (F) or in blood, skeletal muscle, liver, and kidney (G). One-way ANOVA compares concentration in knee to other tissues over time. Combination of two experiments, n = 4 to 5 dosed at 7.5 μCi, 60 nmol, and 7.1 μCi, 57 nmol. All quantified data are means and SD.

To determine the rate of accumulation, peak concentration, and duration of retention in cartilaginous tissues, we evaluated the biodistribution of 14C-CDP-11R in two independent, nine-point, time-course experiments (Fig. 1, F and G). The peptide was observed on the autoradiographs in the knee and IVD at all time points between 5 min and 96 hours. The peak mean concentration occurred at 30 min in both the knee (14.4 ± 3.4 nmol/g) and IVD (8.4 ± 2.1 nmol/g). The biodistribution pattern was similar in rats as in mice (fig. S1)

Comparing the concentration of 14C-CDP-11R in the knee at each time point to the concentration in blood, muscle, and liver revealed that the mean concentration in the knee (Fig. 1F) was significantly higher than in blood, muscle, and liver (Fig. 1G) at all time points (P < 0.05). The mean concentration in the kidney was transiently higher than in the knee from 5 min to 1 hour but was not different between 3 and 96 hours (P > 0.05). Exposure in the kidney was expected because the peptide appeared to be excreted primarily by the kidney and to a lesser extent through the gallbladder into the stool, similar to what we observed with other CDPs (13, 15). These experiments revealed that CDPs can accumulate in the cartilage at high concentration and can be retained longer in cartilage (at least 96 hours) than in other tissues such as blood, muscle, and liver. These CDPs provide a basis for improving the therapeutic window of anti-arthritic drugs through preferential delivery to joints.

Three-dimensional CDP structure is necessary for cartilage accumulation

We used crystallography to determine the three-dimensional CDP structure, which served the dual purposes of providing strong validation of the structural integrity of the peptides and structural insight into their biodistribution patterns. Superposition of three cartilage-accumulating CDPs (CDP-09R, CDP-45R, and CDP-11, the latter a naturally occurring homolog of CDP-11R in which several arginine residues are lysine), as well as a non–cartilage-targeting control, CDP-29, showed a high degree of structural conservation shared by all four CDPs (Fig. 2A). Analysis of electrostatic potential showed that the surfaces of cartilage-accumulating peptides are overwhelmingly positively charged, with only small negatively charged patches, usually associated with the negatively charged C terminus. This finding was in marked contrast to negative control CDP-29. Thus, extensive positive charge may distinguish cartilage-accumulating from non-accumulating CDPs.

Fig. 2 Biochemical characterization of CDPs.

(A) Superposition of four CDP structures solved by x-ray crystallography. The peptide backbone is displayed in a cartoon ribbon, colored either by secondary structure (CDP-11, CDP-09R, and CDP-45R; α-helices: red; β-strands: yellow; and random coil: green), or light blue (CDP-29). Sulfur atoms in disulfide bonds are colored gold. The four CDP structures are shown separately (left column), and as molecular surface representations (middle and right columns), colored by electrostatic potential (positive: blue; negative: red), rotated by 180°. (B and C) Quantification of mean signal with SD of folded (14C-CDP-11R) or linear (14C-CDP-11RS) CDP in cartilage 3 hours (B) and 24 hours (C) after intravenous administration of 14C-CDP-11RS (8.6 μCi, 69.1 nmol) or 14C-CDP-11R (7.5 μCi, 60 nmol) to mice (n = 2).

To test whether the net charge of the peptide primary sequence was sufficient to impart cartilage accumulation or whether the three-dimensional structure was necessary, we evaluated a linearized version of CDP-11R in which the cystine residues were replaced with serine, eliminating the disulfide bonds. Three hours after intravenous administration to mice, the mean concentration of the linear peptide 14C-CDP-11RS was 0.92 ± 0.02 nmol/g in the knee and 0.60 ± 0.16 nmol/g in the IVD, 8.8- and 9.8-fold lower than 14C-CDP-11R (Fig. 2B). At 24 hours, the mean concentration of the linear peptide was 0.19 ± 0.002 nmol/g in the knee and 0.19 ± 0.003 nmol/g in the IVD, 9.5- and 11.5-fold lower than 14C-CDP-11R (Fig. 2C). We concluded that the tertiary structure of the CDP was necessary for joint accumulation and retention in vivo.

CDP-11R accumulates in the proteoglycan matrix of human cartilage explants

To evaluate whether the cartilage-accumulating properties may be relevant to human disease and not simply a rodent-specific finding, we assessed the ability of CDP-11R to accumulate in human cartilage explants. 14C-CDP-11R was retained in human cartilage fragments at a significantly higher concentration than the negative control peptide 14C-CDP-71R (P < 0.0001; Fig. 3A). The mean concentration of 14C-CDP-11R was 97.9 ± 11.1 nmol/g of tissue (uptake ratio, 444.55 ± 107.62) compared to 11.7 ± 3.7 nmol/g (uptake ratio, 6.83 ± 2.02) for 14C-CDP-71R.

Fig. 3 CDP-11R accumulation in human cartilage explants.

(A) Quantification from liquid scintillation counting on cartilage fragments incubated in 10 μM 14C-CDP-11R or 14C-CDP-71R. n = 10 technical replicates. ****P < 0.0001, two-tailed t test. (B) Confocal fluorescent images of human cartilage explants incubated with 10 μM CDP-11R–Cy5.5, CDP-71R–Cy5.5, or Cy5.5 (n = 5) (left, Cy5.5 images). Transmitted light images of toluidine blue (middle) and H&E staining (right). Scale bars, 5 μm.

To visualize CDP localization in the human cartilage explants, we assessed distribution of Cy5.5 stably linked to CDP-11R by an amide bond at the N terminus (CDP-11R–Cy5.5) using confocal microscopy (Fig. 3B). Like 14C-CDP-11R, CDP-11R–Cy5.5 showed strong signal in all human cartilage explants tested, whereas there was little or no fluorescence from the negative control peptide CDP-71R-Cy5.5 or from Cy5.5 alone (Fig. 3B). The CDP-11R–Cy5.5 florescence was observed diffusely throughout the extracellular matrix (ECM), but the lacuna surrounding the chondrocytes appeared to have lower signal. A low, but detectable, amount of intracellular fluorescence was observed in all three conditions and may be attributed to distribution properties of Cy5.5. Hematoxylin and eosin (H&E) and toluidine blue staining demonstrated that the histology and proteoglycan content was similar across cartilage explants from all treatment conditions; however, only CDP-11R–Cy5.5 had an accumulation pattern that overlapped with toluidine blue in the proteoglycan-rich ECM. These data did not allow us to speculate on the cartilage penetration rate or depth that may be observed in human patients, but they did provide evidence that the observed cartilage accumulation was not a rodent-specific phenomenon.

CDP-11R carries a fluorophore to cartilage in vivo

To test the ability of CDP-11R to deliver an active molecule to cartilage in vivo, we investigated whether CDP-11R–Cy5.5 would distribute to cartilage after intravenous administration in mice. Examination of histology sections revealed accumulation of the CDP-11R–Cy5.5 in the same cartilaginous regions as observed for 14C-CDP-11R by QWBA (Fig. 4). These included articular cartilage, meniscus, physeal cartilage, and cartilage of the patella in the knee. Signal was also observed in the tendons, synovium, and periosteum. There was scant signal in the cells of the bone marrow and no appreciable signal in the trabecular or compact bone. The distribution of CDP-11R–Cy5.5 closely approximated the staining pattern of toluidine blue in all cartilaginous tissues, suggesting ECM localization. There was little or no toluidine blue signal in the tendons, synovial membrane, and periosteum, but we did observe signal for CDP-11R–Cy5.5 in these tissues. We did not observe 14C-CDP-11R in these tissues by QWBA, so this distribution pattern may be attributable to Cy5.5. Consistent with human cartilage explant distribution patterns, CDP-11R–Cy5.5 was present in the ECM of the articular cartilage and was largely excluded from chondrocytes. Cy5.5 alone had very low signal in the knee joint, with detectable signal only in cells of the bone marrow. These data demonstrated that CDP-11R delivered Cy5.5 to the cartilage of the knee.

Fig. 4 Joint accumulation of CDP-11R–Cy5.5 in mice.

Representative sections of knee joints from mice that received intravenous CDP-11R–Cy5.5 or Cy5.5 (n = 3). Scale bars, 500 μm. Transmitted light images (10×) of toluidine blue staining and fluorescence images in the Cy5.5 (red) and DAPI (blue) channels (10× and 20×). White box highlights the region that was imaged at 20×. Scale bar is 25 μm in the 20× image.

CDP-11R carries dexamethasone to cartilage in vivo

Having established that cartilage-accumulating CDP-11R could transport an imaging payload, we investigated its ability to deliver a therapeutic payload. Dexamethasone (dex) was selected as the steroid payload for our first CDP-steroid conjugate because of its long history of use in patients with arthritis and preclinical animal models, and due to availability of a water-soluble form [dexamethasone sodium phosphate (dexSP)] for intravenous administration.

Biodistribution was assessed with the stable, cysteinyl-linked CDP-11R–cys-dex conjugate to confirm that the intact conjugate accumulated in cartilage after intravenous administration. Immunocompetent mice were administered CDP-11R, cys-dex, CDP-11R–cys-dex, or vehicle. CDP-11R and/or dex accumulation was assessed in joints collected after 3 hours of in vivo circulation (Fig. 5, A to D). Immunostaining with anti–CDP-11R antibody demonstrated strong positive staining in the articular cartilage of knee joint sections from mice treated with either CDP-11R or CDP-11R–cys-dex; however, there was minimal staining in the cartilage of mice treated with cys-dex or vehicle (Fig. 5A). Immunohistochemistry (IHC) with anti-dex antibody demonstrated strong positive staining in the articular cartilage of mice treated with CDP-11R–cys-dex but minimal cartilage staining in any other treatment group (Fig. 5B). The distribution of staining with an anti–CDP-11R antibody was similar between CDP-11R– and CDP-11R–cys-dex–treated animals. In each case, the peptide was detected in a diffuse pattern throughout the proteoglycan ECM, a result indicating that the addition of the dex cargo did not substantially alter cartilage localization of CDP-11R. The staining did not extend through the full thickness of the articular cartilage evident in the H&E section. Instead, it was restricted to the noncalcified portion of the cartilage, which was separated from the calcified cartilage by the tidemark. The paucity of staining beyond the tidemark may reflect differences in the composition of the cartilage in each region, or it could be an artifact of decalcification in sample processing for IHC. The distribution of staining was similar for both the CDP-11R antibody and the dex antibody in CDP-11R–cys-dex–treated animals, indicating that the conjugate was intact and delivered dex to the cartilage. IHC against CDP-11R showed prominent staining in the chondrocytes in both the CDP-11R and CDP-11R–cys-dex samples. There was, however, faint to moderate staining in chondrocytes in all treatment groups with both antibodies, which suggested that at least some of the cellular staining may have been nonspecific signal. H&E sections confirmed that the cartilage was structurally similar between treatment groups (Fig. 5C), and toluidine blue staining demonstrated similar proteoglycan content (Fig. 5D). Some staining was observed in the bone marrow of CDP-11R and CDP-11R–cys-dex–treated mice; however, bone marrow was not identified as a site of notable accumulation by QWBA or CDP-11R–Cy5.5 microscopy, which suggested an IHC artifact. These data confirmed that conjugation of dex to CDP-11R by a stable linker resulted in delivery of both dex and CDP-11R to cartilage and that free dex did not detectibly accumulate in cartilage of the knee. We thus advanced the CDP-11R–steroid conjugate for further evaluation.

Fig. 5 Biodistribution of CDP-11R–dex in mouse knee.

IHC analysis of CDP-11R–cys-dex accumulation in cartilage after intravenous administration of 100 nmol in mice (n = 4). Rows indicate different treatments; columns indicate different stains or antibodies. Representative knee sections stained with anti–CDP-11R antibody (A), anti-dex antibody (B), toluidine blue (C), and H&E (D). Scale bars, 50 μm.

Development of CDP-11R–dex with a labile linker

We expected dex to be minimally active when conjugated to CDP-11R due to steric hindrance preventing binding with the glucocorticoid receptor. To test a CDP-11R–steroid conjugate that released the steroid at the site of action, we required a CDP-11R–dex conjugate with a labile linker that remained largely intact while in circulation and released the drug in the joints.

We developed CDP-11R–dex with an ester linker that hydrolyzed at a desired rate in plasma. Hydrolysis half-life was examined in vitro using rat and human plasma, with the understanding that hydrolysis may differ somewhat in vivo because of the presence of esterases in organs in addition to blood. The dimethyladipic acid (DMA) linker had a favorable hydrolysis half-life in rat (9.9 hours) and human (22.4 hours) plasma and was advanced to in vivo experiments (Fig. 6A and Table 1).

Fig. 6 Pharmacokinetics of dex release from CDP-11R–DMA–dex in vitro and in rats in vivo.

(A) In vitro hydrolysis of CDP-11R–DMA–dex in PBS and rat and human plasma. Regression line shows the mean of the AUC for dex normalized to the AUC for TAA (internal standard) with individual data points plotted at each time point (n = 3). (B to D) Pharmacokinetic and pharmacodynamic analysis of CDP-11R–DMA–dex administered at 2 mg/kg intravenously to rats. Tissue weight of spleen (B) or thymus (C). (D) Quantification of free dex or CDP-bound dex in plasma in rats. Data presented as means with SD (n = 4 per time point).

Table 1 In vitro hydrolysis.

View this table:

To determine whether the labile linker released functional dex in vivo, we assessed thymus and spleen weights in rats that were dosed with CDP-11R–DMA–dex. The rationale was that dex remained in its active form in plasma; therefore, dex that reached circulation, from either linker release in the blood or efflux of released drug from the joint, would reduce spleen and thymus weights because these organs are exquisitely sensitive to circulating dex. As anticipated, the weight of both the spleen and thymus decreased in animals that received CDP-11R–DMA–dex, similar to dex alone (Fig. 6, B and C, and tables S1 and S2).

Although the pharmacodynamic response of thymus and spleen atrophy conveyed that the dex payload in this construct may not be ideal for further development, we took advantage of the fact that dex could be dosed intravenously and remained in circulation in its active form to assess whether the amount of free dex in plasma was lower when the drug was delivered as a CDP conjugate versus a free drug. Free and CDP-bound dex were quantified by liquid chromatography/mass spectrometry (LC/MS) in the plasma of female Lewis rats 15 min to 24 hours after intravenous administration of CDP-11R–DMA–dex or dexSP. Quantification of total dex (free plus CDP bound) showed that an equal amount of dex was administered as the free or CDP-conjugated form; however, the total exposure, maximum exposure, and time of peak exposure differed between the two forms. Free dex from dexSP peaked at 30 min with a Cmax of 224 ng/ml and area under the curve (AUC) of 1060 hours·ng/ml, compared to a maximum of 71 ng/ml at 6 hours and AUC of 700 hours·ng/ml for CDP-11R–DMA–dex (Fig. 6D, Table 2, and table S3). The in vivo plasma half-life of CDP-11R–DMA–dex was inferred from the quantification of CDP-bound dex as 1.5 hours.

Table 2 Plasma pharmacokinetic parameters.

View this table:

These experiments showed that a dex-CDP conjugate had favorable biodistribution and reduced systemic exposure. However, because of the potency and lack of metabolism to inactive metabolites, dex released from CDP-11R caused on-target pharmacodynamic responses that would be dose-limiting due to systemic steroid exposure when used in the long-term care of human patients.

CDP-11R delivers triamcinolone acetonide to cartilage in vivo

To demonstrate efficacy at the joint and reduced potential toxicity when using the CDP conjugate, we switched the active payload to triamcinolone acetonide (TAA). The higher clearance of TAA, compared to dex, may mitigate the effect of any systemic exposure of released drug (16, 17).

Biodistribution of CDP-11R–14C-cys–TAA or 14C-cys–TAA was assessed 3 and 24 hours after intravenous administration. QWBA demonstrated that the CDP-11R–14C-cys–TAA conjugate had a similar biodistribution pattern to 14C-CDP-11R, with accumulation in cartilage (Fig. 7A); in contrast, unconjugated 14C-cys–TAA had minimal observable cartilage signal (Fig. 7B). Quantitative analysis of tissues showed that, at the 3-hour time point, CDP-11R–14C-cys–TAA accumulated at 9.7 ± 2.2 nmol/g in the knee and at 5.6 ± 2.6 nmol/g in the IVD (Fig. 7C). The mean concentration of CDP-11R–14C-cys–TAA was about 11-fold higher than 14C-cys–TAA in the knee and 33-fold higher in the IVD (Fig. 7C). At 24 hours, CDP-11R–14C-cys–TAA concentration was 0.5 ± 0.1 nmol/g in the knee and 0.9 ± 0.3 nmol/g in the IVD, both significantly higher than 14C-cys–TAA (P < 0.0001; Fig. 7D).

Fig. 7 Biodistribution of CDP-11R–TAA in mice.

(A and B) Representative autoradiograph images of knee and IVD 3 hours after intravenous administration of 100 nmol CDP-11R–14C-cys–TAA (A) or 14C-cys–TAA (B) to mice. Scale bars, 1 mm. (C and D) Quantification of signal in the knee and IVD 3 hours (C) and 24 hours (D) after administration. n = 6. ****P < 0.0001, two-tailed t test.

CDP-11R–DMA–TAA reduces inflammation in arthritic joints and limits systemic steroid exposure

We then investigated the ability of CDP-11R with the DMA linker to deliver TAA to arthritic joints and reduce inflammation in a preclinical model of arthritis. We selected the collagen-induced arthritis (CIA) model in rats because it is widely considered to be the gold standard among rodent models of RA due to its histologic similarity to the human disease and its track record of predicting drug efficacy in the clinic (1821). Inoculation of rats with collagen induced joint inflammation, evidenced by an increase in ankle diameter from 6.83 ± 0.14 mm at baseline to 7.45 ± 0.15 mm at treatment enrollment. Five rats per treatment group were given intravenous doses of CDP-11R–DMA–TAA (0, 58, 115, 230, or 460 nmol/kg) for five consecutive days. The 460 and 230 nmol/kg doses showed a significant reduction in ankle diameter compared to the vehicle group, from 7.75 ± 0.28 mm (vehicle) to 6.92 ± 0.10 mm (460 nmol/kg; P = 0.0048) or 7.11 ± 0.25 mm (230 nmol/kg: P = 0.0163), 24 hours after the first dose of CDP-11R–DMA–TAA (Fig. 8, A to C, and table S4). The difference from vehicle persisted as the study progressed. In addition, at the 460 mg/kg dose, ankle diameter was reduced compared to inflammation at initiation of treatment (day 0) on each progressive day of the study, indicating complete elimination of inflammation (linear regression slope −0.1825, significantly nonzero, P < 0.0001). The rats dosed with 115 nmol/kg showed reduced ankle diameter compared to vehicle, which became significant after four doses (P = 0.00359).

Fig. 8 Reduction in rat ankle joint inflammation by CDP-11R–DMA–TAA.

Response to CDP-11R–DMA–TAA in the rat CIA model of arthritis. (A) Average ankle diameter with SEM. n = 5 rats. ****P < 0.0001, ***P = 0.0003, and *P = 0.036, repeated-measures, two-way ANOVA for change in diameter over time. (B and C) Representative images of ankles on treatment day 4 from the vehicle (B) or CDP-11R–DMA–TAA (460 nmol/kg) (C) groups. (D and E) Tissue weight for spleen (D) and thymus (E) 3 hours after the final treatment. No significant difference in tissue weight observed by two-tailed t test. (F and G) Effect of intravenous TAA administration on mean weight of spleen (F) and thymus (G) tissues 3 hours after five consecutive days of dosing in healthy rats (n = 5). *P = 0.011, ***P < 0.0003, and ****P < 0.0001, two-tailed t test.

Thymus and spleen weights were evaluated 3 hours after the fifth dose of CDP-11R–DMA–TAA. No difference in tissue weight was observed for any of the treatment groups compared to the vehicle control, suggesting that insufficient free TAA reached systemic circulation to induce atrophy in these sensitive tissues (Fig. 8, D and E). In contrast, when 230 or 460 nmol/kg of unconjugated TAA was administered by the same regimen to non-arthritic rats, we observed significant atrophy of the spleen (P < 0.0005) and thymus (P < 0.0001) compared to vehicle-treated rats (Fig. 8, F and G). Together, these data demonstrate that joint targeting CDP-11R can deliver an effective glucocorticoid to the joint, mitigating disease progression and sparing tissues susceptible to steroid toxicity.


Development of a safe and effective targeting agent that can deliver a variety of systemically administered therapeutics to all joints throughout the body would be groundbreaking for the fields of orthopedics and rheumatology. Here, we demonstrated the potential of cartilage-accumulating CDPs to distribute into joint tissues after systemic administration. Time-course studies demonstrated that CDP-11R rapidly accumulated and can be detected in cartilage 4 days after administration despite clearance from plasma. The biodistribution pattern was consistent across species, and CDP-11R was taken up by human cartilage explants. CDP-11R delivered small molecules with diverse structures to cartilage in multiple joints that are commonly affected by arthritis.

We initially considered that a short plasma half-life may be a challenge in the development of CDPs as drugs, but we observed that the arthrotropism of these peptides was strong enough to drive their accumulation in cartilage even while glomerular clearance reduced their exposure to other tissues. The QWBA time-course experiment demonstrated highly disparate concentration versus time profiles for CDP-11R in blood compared to cartilage. Although the peptide is rapidly eliminated from plasma, the concentration in cartilage demonstrated accumulation for hours and retention for multiple days in both the knees and IVDs. This pharmacokinetic profile may present an advantage with regard to drug delivery because the rapid clearance from plasma could limit exposure of the drug to non-cartilaginous tissues.

The ECM of cartilage is composed of negatively charged proteoglycans arranged in feather-like arrays. The high net positive charge at physiologic pH likely played an important role in the biodistribution of cartilage-accumulating CDPs. Donnan equilibrium provides a biochemical explanation for this process in which the fixed net negative charge of the proteoglycans in the cartilage ECM creates an electrical potential that enhances the uptake of cations into cartilage (2224). In addition, once the positively charged molecules have partitioned into cartilage, electrostatic interactions with the negatively charged sulfate and carboxylic acid groups of proteoglycans can lead to weak and reversible binding (25, 26).

The use of positively charged molecules to access cartilage has been described elsewhere. In the 1970s and 1980s, it was reported that positively charged aminoglycoside antibiotics and bis-quaternary ammonium compounds accumulate in cartilage after systemic administration (2730). Other groups are investigating the use of non-CDP small cationic peptides for drug delivery to avascular, negatively charged tissues (31), and Tanner et al. (32, 33) are investigating the use of ion channel blocking venom peptides in rodent models of RA.

Although the high net positive charge of cartilage-targeting CDPs may be sufficient for the initial partitioning into cartilage, it is likely that additional structural components are responsible for their long residence time in the cartilage matrix. Research has shown that Pf-pep, a small linear cationic peptide with therapeutic potential, diffuses into cartilage when incubated ex vivo, but is not retained there long enough to have a therapeutic effect (25). In our study, the CDP-11RS linearized control peptide demonstrated minimal accumulation in cartilage as compared to the disulfide-bonded form at both early and later time points. The linear peptide was similar in both sequence and charge to CDP-11R; therefore, the structural alteration was likely the cause of the diminished targeting. It is plausible that loss of the knotted conformation made the linear peptide susceptible to proteolysis or disrupted the structurally rigid charge distribution. Alternatively, the compact knotted structure may have permitted superior penetration into cartilage and subsequent interaction with proteoglycan molecules, leading to a longer retention time.

Electrostatic interactions have been effectively leveraged for delivery of cationic drug nanocarriers to cartilage after in vitro incubation or intra-articular administration (26, 3437); our work advances this field by demonstrating that a systemically administered peptide can achieve high concentration in the cartilage of joints commonly affected by arthritis. We were encouraged by the high estimated peak concentration of CDP-11R in the cartilage of the knee (14.4 μM) and IVD (8.4 μM) and the potential to deliver therapeutic concentrations of drugs using these peptides as delivery vehicles. This finding was further supported by our observation that CDP-11R–cys-dex and CDP-11R–14C-cys–TAA accumulated in the cartilage of synovial joints and IVDs throughout the mouse. Long-term steroid use can lead to glaucoma, edema, hypertension, mood and psychological problems, redistribution of fat, cataracts, immunosuppression, and elevated blood glucose (38). Preferential delivery to joints rather than the tissues involved in on-target toxicities may be sufficient to achieve an acceptable therapeutic window. Although chronic systemic use of potent steroids is no longer the clinical standard of care because of pernicious cumulative toxicities, the fact remains that steroids are often effective. Steroid injections into individual joints are standard of care, but they require medical provider administration and are not suitable for some multisite diseases such as RA or ankylosing spondylitis. The current study supports further development of CDP-steroid conjugates while also opening a path for nonsteroid disease-modifying payloads that could not only reduce pain associated with disease but ultimately help repair or restore cartilage. The cost of goods for a CDP-steroid therapy may be greater than a steroid alone, but because the CDPs are small and can be made synthetically, it is anticipated to be more economical than biologics such as antibodies.

Although the experiments described here established proof of principle that CDPs could be used to deliver therapeutic agents to diseased joints, further investigations specific to the payload and indications will be indicated in the course of candidate development. First, our experiments with human cartilage explants established that our CDP was able to accumulate in human cartilage, but we did not determine the rate or depth of penetration. This information may be critical for matching the therapeutic residence time in the joint to rate of accumulation. Our steroid delivery conjugate was envisioned to provide a localized depot of steroid to mediate inflammation in the joints and/or synovium; however, payloads that act directly on chondrocytes may require penetration throughout the cartilage. Second, although we showed that our CDP could carry a fluorophore and two steroids, the pharmacokinetic and biodistribution behavior of other payload conjugates will need to be determined. Last, we have shown efficacy in the rat CIA model without evidence of systemic exposure after 5 days of treatment; however, therapy for patients with multijoint arthritis will likely be more prolonged and response to chronic dosing will need to be assessed in longitudinal studies.

Targeted therapeutics have been considered and advanced primarily for cancer, but we believe that CDPs could play a key role in delivering therapeutics to nonneoplastic tissues. The rapid distribution and substantial retention of CDP-11R in cartilage made this molecule and related CDPs candidates for treating multijoint arthritic diseases. Our study in the rat collagen–induced model of RA demonstrated that targeting TAA to the inflamed joint slowed and reversed disease progression without inducing atrophy in the thymus and spleen, tissues that are adversely affected by glucocorticoid therapies. Peptide-drug conjugates developed using this approach may enable meaningful improvements in human and veterinary health.


Study design

The objective of this work was to characterize the arthrotropism of select CDPs and develop a joint-targeted therapeutic agent. Experiments were performed to evaluate biodistribution, pharmacokinetics, and anti-inflammatory efficacy of CDPs or CDP conjugates. Biodistribution was determined in two to six female athymic or C57Bl6 mice. Two experienced researchers verified region of interest (ROI) selection for quantification of signal. Five to six mice were used for statistical comparisons. In vitro hydrolysis experiments were performed in two experiments. Pharmacokinetic evaluation of drug release was performed once using four healthy rats per randomly assigned group. The data analyst was blinded to treatment identity. The rat CIA efficacy study assigned five rats per treatment group at random by rolling enrollment once a predetermined joint measurement was reached. The researcher who conducted and analyzed the experiment was blind to treatment identity.

Production and crystallization of CDPs

CDPs were recombinantly expressed using our custom lentiviral-based mammalian protein expression system described previously (39). Protein sequences are in Table 3; sequence confirmation analysis is in table S5. Each sequence was reverse-translated using human codons and cloned downstream of human siderocalin into our modified pCVL lentiviral vector (40). Lentivirus was produced by transient cotransfection with psPAX2 and pMMD2G of 293T cell and used to transduce FreeStyle 293-F cells (Thermo Fisher Scientific). Secreted fusion protein was extracted from the medium using immobilized metal-affinity chromatography [nickel–nitrilotriacetic acid (Ni-NTA)] and cleaved from siderocalin using an in-house produced tobacco etch virus (TEV) protease under native conditions. CDPs were purified by reversed-phase high-performance LC (rpHPLC). The peptides were aliquoted, lyophilized, and submitted for amino acid analysis (AAA Service Laboratory Inc.).

Table 3 Peptide sequences.

View this table:

Crystallization and crystallography

Crystallization screening was performed at room temperature (RT) by vapor diffusion, with 1:1 protein:reservoir solution sitting drops using submicroliter robotics (TTP LabTech mosquito). Crystals were harvested from screening plates and cryopreserved. Diffraction data were collected from single crystals using a Rigaku MicroMax-007 HF home source. For sulfur single-wavelength anomalous diffraction (sSAD) phasing, Bijvoet pair measurement was optimized by collecting data through 5° wedges with alternating phi rotations of 180°, in 1° oscillations. Data were reduced and scaled with HKL2000 (41). Initial phases for the CDP-29 structure were determined by sSAD (42), the sulfur substructures defined with SHELX (43), or by molecular replacement using PHASER (44) in the CCP4 program suite (44), and using the CDP-29 or CDP-11 structures as search models. Iterative cycles of model building and refinement were performed with COOT (45) and REFMAC (46); structure validation was performed with MolProbity (47). The structures have a maximum likelihood root mean square deviation of 0.18 Å, determined by Theseus (48). Further experimental details are in table S6. Analysis of peptide isoelectric point for comparison to crystal structure is in table S7. Coordinates have been deposited in the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank (49).

Radiolabeling of CDPs

Peptides were exhaustively methylated using the procedures of Jentoft and Dearborn (50, 51). Peptide, 14C-formaldehyde (57 mCi/mmol; Pharmaron), and sodium cyanoborohydride were reacted at RT overnight. 14C-CDP was purified using a Strata-X reversed-phase cartridge (Phenomenex). Dried samples were stored at −18°C. Radiochemical purities and HPLC traces are in the Supplementary Materials (table S8 and fig. S2, A to E).

Fluorophore conjugation

Sulfo-Cy5.5 N-hydroxysuccinimide (NHS) ester (Click Chemistry Tools) was dissolved in dimethyl sulfoxide (DMSO) and added to peptide in 50 mM sodium bicarbonate (pH 8). The reaction was purified by HPLC, and Cy5.5-CDP fractions were analyzed by LC/MS. HPLC eluent was quantitated on a NanoDrop spectrometer (Thermo Fisher Scientific) at 678 nm using an extinction coefficient of 190,000.

Synthesis of CDP-steroid conjugates

Detailed chemistry methods, chemical structures, and HPLC traces showing product purity are in the Supplementary Materials and figs. S3 and S4. The corresponding dex conjugates were made using a similar method.

CDP-11R–Cys–TAA conjugation. CDP-steroid conjugate was synthesized on the basis of the method of Ma et al. (52). Methane sulfonyl chloride and TAA in anhydrous pyridine were reacted for 30 min; workup and solvent were removed under reduced pressure. TAA mesylate in dimethylformamide (DMF) was added to tert-butyloxycarbonly (Boc)-l-cysteine and cesium carbonate in DMF at RT for 2.5 hours, filtered, and purified by rpHPLC. Product fractions were pooled, frozen, and lyophilized. 1-Ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and NHS were added to TAA–Boc-cysteine in DMF, stirred at RT overnight, and purified by HPLC. TAA–Boc-cysteine–NHS ester was added to CDP-11R and n-methylmorpholine (NMM) in DMSO, stirred at RT overnight, and purified by HPLC. TAA–Boc-cysteine–CDP-11R was dissolved in trifluoroacetic acid (TFA). Acetonitrile (ACN) and water 1:1 were added, and the solution was lyophilized. The product was 14C-labeled using reductive methylation.

14C-cys–TAA synthesis. TAA–Boc-cysteine was dissolved in methanol, hydrochloric acid was added, and the reaction was stirred at RT overnight. The solvent was removed under reduced pressure, and the residue was dissolved in DMSO and purified by HPLC. The product was deprotected and 14C-labeled using reductive methylation.

CDP-11R–DMA–TAA conjugation. TAA, DMA, EDC, and 4-dimethylaminopyridine were dissolved in dimethyl carbonate and reacted at RT overnight. The solvent was removed under reduced pressure, and the residue was purified by HPLC. DMF was added to TAA-DMA acid; EDC and NHS then reacted at RT for 1.5 hours. Additional NHS and EDC were added and reacted at RT overnight and then purified by HPLC. TAA-DMA-NHS ester was added to CDP-11R and NMM in DMSO, reacted at RT for 60 hours, and purified by HPLC.

Ex vivo accumulation of CDPs in cartilage

Human cartilage from postmortem donors with no history of joint disease or arthritis was purchased through Articular Engineering. Cartilage slices from the knees of two female donors, aged 47 and 65, were received as fresh, live tissue in medium with 10% fetal bovine serum within 48 hours of death and used immediately. Biopsy punches (3 mm) were taken through the thickest part of the slices, resulting in samples 1 to 2 mm thick for ex vivo incubation. Samples were incubated in 100 μl of serum-free Dulbecco’s modified Eagle’s medium (DMEM) with 10 μM CDP-11R–Cy5.5, CDP-71R–Cy5.5, Cy5.5 acid, 14C-CDP-11R, or 14C-CDP-71R in phosphate-buffered saline (PBS) for 24 hours on an orbital shaker.

After fluorophore incubation, samples were washed 3 × 10 min in Hanks’ balanced salt solution (HBSS; Fisher Scientific), pH 7.4, stained for 20 min with Hoechst 33332 (5 μg/ml; Sigma-Aldrich) in HBSS. Samples were washed 2 × 10 min in HBSS, transferred to a 35-mm imaging dish (Ibidi) filled with HBSS, and kept on ice. After imaging, samples were fixed in 10% neutral buffered formalin (NBF; Fisher Scientific).

After radioactive incubation, samples were transferred to a new 96-well plate, washed in tris-buffered saline with Tween 20 (TBST) 3 × 10 min, and then dissolved in 100 μl of collagenase II (10 mg/ml; Sigma-Aldrich) at RT for 48 hours on an orbital shaker. The entire sample was added to 4 ml of Ultima Gold scintillation fluid (PerkinElmer) and counted on a Packard Tri-Carb scintillation counter (PerkinElmer).

In vitro hydrolysis assay

CDP-11R–DMA–dex was reconstituted in DMSO; diluted in PBS, human, or rat plasma; and rocked at 37°C. At each time point, 100 μl of hydrolysis solution was added to 1 ml of ACN with standard (2.5 μg of TAA) at 4°C. Samples were vortexed and centrifuged, and supernatant was transferred to a 96-well plate. The pellet was resuspended in 500 μl of ACN and repelleted, and the supernatant was added to the same well in the block.

All samples were dried under a nitrogen stream and reconstituted in 110 μl, 45:55 ACN:citrate 10 mM (pH 5.5). The block was shaken at 7000 rpm and then centrifuged. Samples were transferred to a 96-well plate and centrifuged at 6500g for 15 min immediately before LC/MS. Analysis was performed on an Agilent 1260 Infinity series HPLC with in-line 6120 single quad MS using a 25 to 65% ACN:water (+0.1% TFA) gradient on an SB-C18 column (InfinityLab). Dex was quantified by integration of analyte and internal standard peaks of the total ion count. Percent hydrolyzed was calculated assuming full drug release was achieved at the end of the assay.

Animal research

All rodents were maintained in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals with approval from the Fred Hutchinson Cancer Research Center Institutional Animal Care and Use Committee (protocols 50808 and 50868). Female athymic nude mice and Lewis rats (Envigo) were used in study when 8 to 10 weeks old. Female C57BL6 mice (The Jackson Laboratory) were used when 12 to 14 weeks old. All rodents were group-housed with unrestricted mobility and free access to food and water.

In vivo biodistribution

14C- or fluorophore-labeled CDPs, dex- or TAA-conjugated CDPs, and appropriate fluorophore or glucocorticoid controls were administered intravenously as a 100-μl bolus. 14C-labeled peptides were administered in PBS, peptide-steroid conjugates, and controls in 5% DMSO in water, and CDP-11R–Cy5.5 and Cy5.5 acid in 2.5% DMSO in PBS. Each compound was allowed to circulate before animals were humanely euthanized, and tissues were analyzed by QWBA, microscopy, or IHC.

Quantitative whole-body autoradiography

QWBA experiments were conducted based on industry standards (53). Frozen mice were embedded in 2% carboxymethylcellulose, and three to four 40-μm sagittal sections were cut on an H/I Bright 8250 Cryostat (Hacker Instruments). Radioactive control guides of 14C-glycine [American Radiolabeled Chemicals (ARC)] were drilled into each block. Sections were freeze-dried at −20°C for 48 to 72 hours, mounted on sturdy paper, and covered with cellophane (Reynolds Food Service Film). Sections were exposed to phosphor imager plates (Raytest) with a standard curve (ARC) for 7 days and scanned on a Raytest CR-35. Radioactivity signal in tissues was quantified from signal intensity using AIDA Image Analyzer v 5.1 Whole Body Autoradiography Professional software (Raytest). ROIs were manually drawn around tissues in each section; tissues in multiple sections were averaged to yield one value per tissue per mouse. Tissue (nanomoles per gram) was interpolated from the standard curve. Detailed methods are in the Supplementary Materials and fig. S5.

In vivo pharmacokinetic study

Rats received either dexSP (0.26 mg/kg; Henry Schein) or CDP-11R–DMA–dex (2.5 mg/kg), equimolar for dex, intravenously. Free dex was quantified in plasma at 0.25, 0.5, 1, 2, 3, 4, 6, 12, and 24 hours (n = 4 rats per time point). An untreated group was included as a negative control. At euthanasia, blood was collected in BD K2EDTA vacutainer tubes (Fisher Scientific), and total body weight and organ weights (thymus and spleen) were recorded. Frozen plasma was shipped to Alturas Analytics for quantitation.

Quantitation of dex in plasma by HPLC/MS/MS

Dex was extracted from plasma samples in duplicate using ACN. Each sample (25 μl) was aliquoted to a 96-well plate with 1:1 dex-D4 in water/methanol (internal standard, 10 ng/ml). The samples were centrifuged at 3000 rpm and vortexed for 5 min each. ACN (500 μl) was added, and samples were vortexed and centrifuged at 3000 rpm for 5 min each. Supernatant was removed and evaporated and then reconstituted with 100 μl of water/ACN 1:1, vortexed, and centrifuged at 3000 rpm for 5 min each.

A second aliquot of each plasma sample underwent ex vivo hydrolysis to force intact conjugate to release dex. Ten microliters (10 U) of porcine esterase (Sigma-Aldrich) was added to 25 μl of plasma and incubated ~22 hours at 37°C with light shaking. Released dex was extracted using the method above.

Extracts were injected onto an HPLC/MS/MS triple quadrupole mass spectrometer (Sciex API 6500). A Phenomenex Luna C18 column (5 cm × 2.0 mm, 3 μm) was used to separate dex and dex-D4 from interfering compounds. The lower limit of quantitation was 0.4 ng/ml. A standard curve was generated by spiking dex into plasma at 0.4, 0.8, 2, 10, 50, 100, 350, or 400 ng/ml, extracted, and analyzed in duplicate. The average accuracies and coefficients of variation of the eight concentrations were >97% and <8% respectively.

Pharmacokinetic data analysis

To estimate the amount of intact CDP-11R–DMA–dex at each time point, free and total dex concentrations in nanograms per milliliter were converted to nanomolar (nM) concentrations using the molecular weight of dex (392.46 g/mol). The nM concentration of intact CDP-11R–DMA–dex was then calculated as follows: (total dex nM concentration) – (free dex nM concentration). WinNonlin Phoenix v6.4 was used to generate plasma pharmacokinetic parameters using standard noncompartmental analysis (54).


CIA was initiated in 8- to 10-week-old female Lewis rats. Bovine type 2 collagen (400 μg; Chondrex Inc.) emulsified in incomplete Freund’s adjuvant (Sigma-Aldrich) was injected intradermally on day 0 and 100 μg injected intradermally on day 7. Ankle diameter was measured daily under isoflurane anesthesia beginning on day 10. Animals were randomly placed into treatment groups when ankle diameter exceeded 7.20 mm. Vehicle (5% DMSO in water) or CDP-11R–DMA–TAA was administered at 2.5 ml/kg intravenously to anesthetized rats for five consecutive days. Rats were euthanized 3 hours after the fifth dose; thymus and spleen were collected and weighed.


Human cartilage explants and mouse knee tissues were fixed in 10% NBF for 24 hours. After fixation, knee sections were decalcified with Formical-4 (StatLab Medical Products) for 6 hours with agitation.

All tissues were embedded in paraffin and sectioned at 4 μm. Serial sections were acquired for IHC, 4′,6′-diamidino-2-phenylindole (DAPI), H&E, and toluidine blue. Slides were deparaffinized and washed in water for staining. DAPI staining was performed using Vectashield antifade mounting medium with DAPI (VWR). H&E staining was performed using standard procedures with Harris hematoxylin and bluing solution (VWR) and eosin Y, 1% alcohol (Fisher Scientific). Toluidine blue staining was performed using standard procedures with 0.04% toluidine blue (Fisher Scientific) for 10 min and 0.2% Fast Green FCF (Fisher Scientific) counterstain. Slides were mounted with MM24 synthetic mounting medium (Leica).

For IHC with anti-dex antibody (ab35000, Abcam) and anti–CDP-11R antibody (Fred Hutchinson Cancer Research Center), antigen retrieval was performed using Protease 3 endopeptidase (Roche) at 35°C for 32 min. Primary antibodies were diluted 1:200 (anti-dex antibody) and 1:100 (anti–CDP11R antibody) in casein diluent (Roche), incubated at 37°C for 28 min. Antigens were detected using anti-rabbit-HQ (hapten, Roche), anti-HQ-HRP (horseradish peroxidase) (Roche) at 37°C for 16 min, and ChromoMap 3,3′-diaminobenzidine substrate (Roche). A hematoxylin (Roche) and bluing (Roche) counterstain was applied for 8 min at RT, and slides were dehydrated, cleared, and mounted.

Anti-CDP11R polyclonal antibody generation

CDP-11R was conjugated to keyhole limpet hemocyanin (KLH) (Imject mcKLH, Thermo Fisher Scientific) following the manufacturer’s protocol. CDP-11R–KLH was emulsified with Freund’s complete adjuvant (Sigma-Aldrich) and administered to New Zealand white rabbits following a standard schedule (R&R Research LLC). The polyclonal antibody was purified from the serum using a HiTrap MabSelect cartridge (GE Healthcare). Western blot confirmed antibody specificity.


Human cartilage fluorescent images were acquired on a Zeiss LSM 780 confocal microscope in the DAPI (405-nm excitation, 410- to 460-nm detection) and far-red channels (640-nm excitation, 650- to 710-nm detection). The same imaging parameters were used for all samples. ZEN software was used for image acquisition, and images were processed in ImageJ.

The CDP-11R–Cy5.5 images were acquired using a TissueFAXS automated microscope with Zeiss 10× and 20× objectives. An X-Cite XLED light source was used for excitation with 381 to 403 nm for DAPI and 626 to 644 nm for Cy5.5. Images were processed in ImageJ. Adjustments to brightness and contrast were made uniformly. Transmitted light imaging was used for H&E- and toluidine blue–stained sections, which were scanned using Aperio Scanscope AT with 20×/0.75 NA (numerical aperture) Plan Apo objective. Images were acquired at ×20 magnification using HALO image analysis software (Indica Labs).

Statistical analysis

Statistics were performed using GraphPad Prism v7.0 software. Shapiro-Wilk test was used to test normality of each data set analyzed. QWBA time course: Unpaired two-tailed t test using Holm-Sidak method to adjust for multiple comparisons was used to compare the mean concentration in the knee to each tissue at each time point (n = 4 to 5). Steroid delivery QWBA: Unpaired two-tailed t test was used to compare the mean concentration of TAA-control versus CDP-11R–cys–TAA in the knee at 3 and 24 hours and the IVD at 24 hours (n = 6). Human cartilage explant study: Unpaired two-tailed t test was used to compare the mean concentration of peptide (n = 10). Rat CIA efficacy study: Repeated-measures two-way analysis of variance (ANOVA) with Dunnett’s test of significance compared mean ankle diameter between treatments at each time point. Linear regression of slope analysis was performed on ankle diameter measurements.



Fig. S1. QWBA of CDP-11R in rats.

Fig. S2. HPLC analysis of 14C-CDPs.

Fig. S3. Chemical structures of CDP constructs.

Fig. S4. HPLC analysis of CDP constructs.

Fig. S5. Standard curve calibration of radioactive standards.

Table S1. Individual subject data for spleen weight in the CDP-11R–DMA–dex pharmacokinetic study.

Table S2. Individual subject data for thymus weight in the CDP-11R–DMA–dex pharmacokinetic study.

Table S3. Individual subject data for dex quantification in plasma.

Table S4. Individual subject data for the CDP-11R–DMA–TAA rat CIA efficacy study.

Table S5. Validation of recombinant protein amino acid composition and mass.

Table S6. Crystallization conditions.

Table S7. Predicted isoelectric point of CDPs.

Table S8. Radiochemical purities of 14C-CDPs by HPLC.


Acknowledgments: We thank R. Steele, W. Johnsen, W. de van der Schueren, K. Pilat, and S. Turnbaugh from the Molecular Design and Therapeutics Core for producing and analyzing the CDPs that were integral to all of these experiments; A. Watson and C. Brock for assistance with crystallization of the CDPs; C. Christianson from Alturas Analytics for LC/MS/MS analysis of dex concentration in plasma; J. Vasquez and D. McDonald from Fred Hutch Scientific Imaging for assistance with microscopy; S. Pillai and A. Koehne for providing histopathology expertise for analysis of signal localization in histological sections; and A. Bandaranayake and J. Carter from Fred Hutch Optide Drug Discovery for intellectual input and scientific guidance. Funding: The project was supported by NIH grants R01CA135491-07 (to J.M.O.), Project Violet, the Wissner-Slivka Foundation, the Kismet Foundation, the Sarah M. Hughes Foundation, Strong4Sam, Yahn Bernier and Beth McCaw, Len and Norma Klorfine, Anne Croco, Pocket Full of Hope, and Blaze Bioscience. Author contributions: M.L.C.S., E.J.G., and J.M.O. designed the research. M.L.C.S., E.J.G., G.H., C.Y., F.P., E.N., R.R., and M.M.G. performed the experiments. M.L.C.S., E.J.G., G.H., M.-Y.B., K.B.-B., N.W.N., D.M.M., E.N., R.R., M.M.G., C.M., A.D.S., A.J.M., C.E.C., R.K.S., J.A.S., and J.M.O. contributed to data analysis and interpretation. M.L.C.S. and E.J.G. prepared the figures and wrote the manuscript. G.H., C.Y., F.P., and R.R. contributed to writing the methods. All authors contributed to review of the manuscript. Competing interests: Blaze Bioscience retains intellectual property rights to the CDPs used in this manuscript. J.M.O. is a founder and shareholder of Blaze Bioscience Inc. N.W.N. and D.M.M. are employees of Blaze Bioscience Inc. K.B.-B. is a paid consultant. J.M.O., A.D.S., E.J.G., R.K.S., C.M., C.E.C., and N.W.N. are inventors on patent application PCT/US2016/051166 (“Cartilage-homing peptides”), and G.H., A.J.M., J.A.S., C.Y., C.M., M.L.C.S., and J.M.O. are inventors on patent application PCT/US2019/028406 “Conjugates of Cartilage-Homing Peptides”. F.P., M.-Y.B., E.N., R.R., and M.M.G. have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials. Protein crystal structures have been deposited in the Protein Data Bank (, accession numbers: 6ATM.pdb, 6ATY.pdb, 6AY7.pdb, and 6AY8.pdb. Requests for additional information and/or materials related to this study should be addressed to the corresponding author and may be fulfilled through a material transfer agreement.

Stay Connected to Science Translational Medicine

Navigate This Article