Mast Cell Degranulation and Fibroblast Activation in the Morphine-induced Spinal Mass: Role of Mas-related G Protein-coupled Receptor Signaling

The pivotal role played by spinal systems in the encoding of information relevant to the expression of pain is nowhere more evident than the selective attenuation of the response to strong and/or tissue-damaging stimuli induced by intrathecal opioids. This action is mediated by opioid receptors on nociceptive primary afferents and second order dorsal horn neurons.^1,2  Development of chronic implantable spinal delivery systems allowed continuous intrathecal opioid infusion as a chronic pain therapy.³ 

The observation that patients receiving intrathecal opioids displayed neurologic signs secondary to local spinal cord compression was not reported in humans until 1991,^4,5  and then recapitulated in dogs,⁶  sheep,⁷  and guinea pigs.⁸  Preclinical work emphasized that this compression resulted from a fibroblast-rich mass in a collagen matrix arising from the dura-arachnoid adjacent to the delivery site.^6,8,9 

The intrathecal mass has several defining properties: (1) it is produced by several opioids (e.g., morphine, hydromorphone), but not others (e.g., fentanyl and alfentanil)^10,11 ; and (2) formation in dogs is not prevented by opioid antagonism. These observations support the argument that these effects parallel the pharmacology of mast cell degranulation and are not mediated by an opioid receptor.¹²  Meningeal mast cells are indeed degranulated in a naloxone-independent fashion by morphine.¹¹  The role of mast cell degranulation in intrathecal morphine-induced mass formation is supported by the observation that mast cell stabilizers could reduce the incidence or size of the intrathecal mass in dog.⁹  Accordingly, we hypothesized that the mass is produced by degranulation of meningeal mast cells, releasing agents that stimulate fibroblastic activity and collagen deposition,⁸  or by a direct effect upon fibroblast proliferation and migration.¹³  Degranulation of mast cells by morphine is mediated by a family of Mas-related G protein-coupled receptors.^14,15  This leads to the hypothesis that agents activating Mas-related genes will produce mast cell degranulation and intrathecal masses, while conversely, this effect will be absent in opiates that neither activate Mas-related genes nor degranulate mast cells.

The present studies address three questions using the guinea pig intrathecal model.⁸  (1) Does opioid antagonism prevent morphine-evoked masses in a second species? (2) Do opioid agonists showing reduced mast cell degranulation or fibroblast activation show reduced mass formation at equianalgesic concentrations? (3) Does the mast cell degranulating effect, the effects upon fibroblast proliferation, and the ability to activate Mas-related genes covary with the propensity of a specific opioid to yield an intrathecal mass? Recent work demonstrates that the mu-opioid agonist morphine activates Mas-related genes and results in degranulation of mast cells.¹⁵  Conversely, PZM21, developed as a biased μ-opioid ligand, fails to activate Mas-related genes and, as we show here, will not degranulate mast cells or produce a spinal mass. Alternatively, mu-opioid ligands such as the dermorphin-derived peptides (2’,6’-dimethyl tyrosine-[Tyr-D-Arg-Phe-Lys-NH2], abbreviated as DMT-DALDA) are extremely potent after intrathecal delivery,^16,17  and while they degranulate mast cells, we show here that the low concentrations required to produce mu-opiate receptor–mediated analgesia fails to produce a mass in vivo.^18,19 

These studies were carried out according to a protocol approved by the Institutional Animal Care and Use Committee at the University of California, San Diego, California.

Adult male Hartley guinea pigs (250 to 275 g) were purchased from Charles River Labs (USA). Animals were pair-housed in standard cages and maintained on a 12:12-h light–dark cycle. After a minimum of 2 days of acclimation, animals were prepared with chronic lumbar intrathecal catheters. For continuous infusion, the catheter was connected to subcutaneously implanted osmotic minipumps (Alzet model 2002; DURECT Corporation, USA). Guinea pig chow and water were freely available. All procedures and testing were conducted during the daylight cycle.

Guinea pigs were randomly selected for study and prepared with preloaded intrathecal catheters and pumps. Before initiation of drug delivery, baseline behavioral and testing data were taken. At selected times after the initiation of infusion of the test or control article, data were collected again. All nociceptive and behavioral assessments were made with observers blinded as to the pump contents. Behavioral data review and analysis were performed after the completion of the histologic assessments.

• (1) Acute Thermal Escape. To assess hind paw thermal escape, a Hargreaves-type hind paw thermal stimulator system was used²⁰  (Torrey Pines Instruments, USA). This system directed a focused light beam on the plantar surface of the paw, through a glass plate upon which the guinea pig stood. Surface temperature was maintained at 28°C. The guinea pigs were placed in the thermal escape box and allowed to acclimate for approximately 30 min before testing. A brisk withdrawal of the paw was taken as the response. Lack of a response within 20 s was cause to terminate the test and assign the score of “20 s.” A latency measurement was taken for the right and left hind paws and averaged for the analysis datum. Latency assessments were then made at intervals before and after initiation of drug delivery.

• (2) Motor Assessment. Motor assessments included: righting response, symmetrical ambulation, and hind paw proprioceptive function (return of paw to normal position after placing the dorsum of the paw on a horizontal surface). All assessments were noted as “present,” “absent,” “within normal limits,” or “abnormal.”

• (3) Behavioral Assessment. General behavioral assessments were made during each period of observation and specifically included: corneal reflex, pinna reflex, scratching and motor assessment. For analysis, the mean ± SEM of the thermal escape latencies were plotted versus time. To facilitate comparisons between groups, data for each animal were converted to area under the curve where the escape latency data were converted to percent of the maximum possible effect (%MPE = [Test value T_x − T₀ value]/[Cutoff value − 20 s − T₀ value]) × 100. The area under the percent of the maximum possible effect curve was then calculated for each animal. Motor assessment included: righting response, symmetrical ambulation, and hind paw proprioceptive function (return of paw to normal position after placing the dorsum of the paw on a horizontal surface). All assessments were noted as “present,” “absent,” “within normal limits,” or “abnormal.”

Animals were prepared following aseptic precautions with lumbar intrathecal catheters and connected to osmotic minipumps. This procedure and the construction of the associated catheter are explicitly detailed elsewhere.⁸  In brief, the polyurethane intrathecal catheter was constructed from polyurethane (polyurethane 10 0.13-mm inside diameter/0.27-mm outside diameter), and the external component was polyurethane (polyurethane 27 0.40-mm inside diameter/0.70-mm outside diameter) (Instech Laboratories, Inc., USA). An additional short piece of polyurethane tubing (polyurethane 40: 0.60-mm inside diameter/1.0-mm outside diameter) was fitted to allow connection to the osmotic pump. All were joined by a bonding agent (thin instant adhesive gap fill, 0.001–0.003 inches; Scientific Commodities, USA). A stainless steel Teflon-coated stylet (0.003 ± 0.0003 × 73 inches, EZ Grind process; Wytech Industries, Inc., USA) was fed into the catheter lumen. Each catheter was individually packaged, and ethylene oxide sterilized. For catheter placement, guinea pigs were sedated (ketamine: 20 mg/kg, intraperitoneal; xylazine: 2 mg/kg, intraperitoneal) and the back of the neck shaved and surgically prepped. The animal was placed in a rodent stereotaxic head holder and anesthesia maintained by delivery of oxygen and isoflurane (2%) through a fitted face mask. Animals were monitored during the procedure to ensure adequate anesthetic depth. The catheter was inserted through an incision in the atlantooccipital membrane following exposure by incision and blunt dissection and passed 11.5 cm caudally to the T13-L1 spinal segment. For continuous infusion, the catheter was tunneled caudally to a site on the upper back, where a subcutaneous incision was made, and a small pocket was dissected to hold the osmotic mini pump. The catheter was then connected. The wound was sutured in layers. Upon terminating the anesthesia, a subcutaneous bolus of lactated Ringer’s (1 ml/50 g bodyweight) was given between the scapulae. An injection of Rimadyl (5 mg/kg, subcutaneous; Zoetis Inc., USA) was always given for postoperative analgesia. Average duration of the implant procedure was 20 to 30 min.

The following drugs were employed for intrathecal delivery. Morphine was prepared from commercially available, preservative-free morphine sulfate for intrathecal use (morphine sulfate injection, MW: 379 Da; USP 25 mg/ml, preservative free; Hospira, Inc., USA). The μ agonist DMT-DALDA ([H-Dmt-D-Arg-Phe-Lys-NH2; Dmt. 2’,6’-dimethyltyrosine] MW: 981 Da) was synthesized as previously described.¹⁹  The biased μ ligand PZM21 (MW: 362 Da) was synthesized as previously described.¹⁵  All solutions were diluted and prepared using preservative-free sterile saline for injection (0.9% sodium chloride injection, United States Pharmacopeia; Hospira, Inc.). The sodium chloride package insert lists the pH between 4.5 and 7.0.

For the ex vivo studies with mast cells and fibroblasts, morphine, PZM21 and DMT-DALDA, and fentanyl were employed. In addition, to assess Mas-related G protein-coupled receptor-X activation in the PRESTO-Tango assays, additional drugs were employed, including: TAN67 (a potent δ-opioid agonist: R*,S*)-(±)-2-Methyl-4aa-(3-hydroxyphenyl)-1,2,3,4,4a,5,12,12aa-octahydroquinolino [2,3,3-g]isoquinoline dihydrobromide: MW: 387 Da; Sigma; Cat. no. T5824) and TRV 130 (a biased μ-opioid agonist; MW: 344 Da; courtesy Bryan Roth, M.D., Ph.D., Department of Pharmacology, University of North Carolina, Chapel Hill, North Carolina); mitiglinide (MW: 315 Da; Sigma Cat. no. SML0234) serves as positive control for Mas-related gene activation, deoxycholic acid (MW: 393 Da; Sigma Cat. no. D2510) is a partial Mas-related gene agonist, and cromolyn sodium (MW = 468 Da; Sigma Cat. no. C0399) was assessed for its ability to activate Mas-related genes or antagonize the effects of Mas-related gene agonists.

In the current studies, guinea pigs were randomly selected from their group-housed cages to receive infusion pumps prepared to deliver 0.5 µl/h for 14 days of one of the following formulated drugs: vehicle (0.9% NaCl), morphine sulfate (25 mg/ml: 66 nmol/µl), PZM21 (20 µg/µl: 20.4 nmol/µl), or DMT-DALDA (19.6 ng/µl: 20 pmol/µl). To determine the effects of systemic opioid antagonism on spinal mass formation, a second pump was also placed subcutaneously to deliver naltrexone HCl (50 mg/ml: 25 µg/h; approximately 75 µg · kg^-1 · h^-1 for 14 days; Sigma-Aldrich, USA) in a group of animals receiving intrathecal morphine. The concentration of morphine employed in this study was based on previous work in this model.⁸  The concentrations of DMT-DALDA and PZM21 employed in the current studies, which were equiactive to morphine, were determined in preliminary concentration ranging studies with intrathecal infusion in intrathecal catheterized adult male Sprague-Dawley rats using osmotic pumps (unpublished data).

For tissue harvest, guinea pigs were deeply anesthetized with isoflurane anesthetic and given a 1.0 ml intraperitoneal injection of Beuthanasia-D (euthanasia solution consisting of phenytoin/pentobarbital; Intervet/Schering-Plough Animal Health Corp., USA). They were then transcardially perfused with 1 ml/g of body weight of heparinized saline followed by 1 ml/g of body weight of 4% paraformaldehyde in 0.1 M phosphate buffer. Spinal columns were harvested and underwent decalcification for 3 to 5 days at room temperature using Immunocal bone decalcifier (American MasterTech Scientific, Inc., USA). After decalcification was complete, the spinal column was blocked via a template; block A (cervical), B (thoracic), and C1, C2, C3, C4 (lumbar/sacral) placed in tissue cassettes and processed to paraffin. Paraffin blocks were sectioned at 5 to 10 μm using a rotary microtome (Leica 2135, Leica Biosystems, Germany), and sections were slide mounted. Slides were stained with hematoxylin and eosin and Picro-Sirius for collagen. Hematoxylin and eosin staining was performed using Gills Hematoxylin #2 and Eosin Y alcoholic (Sigma Millipore, USA). Hematoxylin and eosin images were captured using an Olympus BX51 microscope (Olympus, USA) and Lumenera Infinity 3S camera with Infinity Analyze software (Lumenera Corporation, Canada). Picro-Sirius collagen staining was performed following the manufacturer’s protocol (American MasterTech Scientific, Inc., USA). Picro-Sirius bright field and polarized birefringence images were captured on a Nikon Eclipse 80i microscope with Nikon DS-Fi1 camera and Digital Sight DS-L2 software system (Nikon Inc., USA). When viewed under polarized light, Picro-Sirius stained sections showed thick (yellow-orange birefringence) and thin (green birefringence) collagen fibers, respectively, of types 1 and 3 collagen.^21,22 

For pathology assessment, all animals submitted for tissue harvest were analyzed. If the catheter penetrated the cord, the presence of the parenchymal catheter was noted, but no score was given. Two quantification approaches were employed, as previously described.⁸  In brief, three reviewers independently, and without knowledge of treatment, first assigned a pathology grade (0 to 4) to each animal based on criteria previously outlined. Second, each reviewer performed a forced ranking of all representative sections from least to most severe and each animal was then assigned the rank given to them by the reviewer. In each case, the mean of the three pathology scores or ranks was assigned to that animal. The score and ranking for each animal were then presented as scatter plots by treatment and used in the subsequent comparisons of treatment effects.

LAD2, a bone marrow-derived human mast cell line,²³  was cultured in StemPro-34 serum-free medium with StemPro-34 Nutrient Supplement (Gibco, USA), l-glutamine (2 mM; Gibco), penicillin (100 U/ml)/streptomycin (100 µg/mL; Gibco), and 100 ng/ml recombinant human stem cell factor (R&D Systems, Inc., USA). Half of the media was replaced weekly by adding an equal volume of fresh StemPro-34 containing 100 ng/ml recombinant human stem cell factor. Well concentrations were maintained between 0.25 to 0.5 × 10⁶ cells/ml to minimize clumping.²⁴  Degranulation was assessed by measuring the activity of β-hexosaminidase in the supernatants²⁵  of 5 × 10⁵ LAD2 cells in 200 µl saline phosphate buffer (0.9% NaCl, 10 mM NaH₂PO₄, 45 mM glucose), incubated for 2 h with the test agents (morphine, DMT-DALDA, and PZM21). For each sample assayed, cell supernatant aliquots (20 μl) were mixed with substrate solution (100 µl: 1 mM 4-methylumbelliferyl-2-acetamide-2-deoxy-β-D-glucopyranoside; EMD Millipore, USA) in 0.1 M sodium citrate buffer (pH 4.5), and were incubated for 2 h at 37°C in the dark. Reaction was stopped by addition of 12 µl of 0.2 M glycine (pH 10.7). Reaction mixtures were excited at 365 nm and measured at 460 nm in a fluorescence plate reader (Gemini EM microplate spectrofluorometer; Molecular Devices, USA). To determine the total cellular content of this enzyme, an equivalent number of cells were lysed with 1% triton X-100 (Sigma Aldrich, USA). Release of β-hexosaminidase was calculated as the percentage of the total enzyme content. Compound 48/80 was used to promote complete mast cell degranulation in an IgE-independent way.²⁶ 

Mouse embryonic fibroblasts were obtained from ATCC (#CRL-2214). Mouse embryonic fibroblasts were cultured in 75-cm² culture flask in 1X Dulbecco’s Modified Eagle Medium supplemented with 10% non–heat inactivated fetal bovine serum, 1% penicillin/streptomycin, 1% sodium pyruvate (100 mM), 1% Glutamax 100X, and 1% minimal essential medium nonessential amino acids 100X (Thermo Fisher Scientific, USA) and incubated at 37°C in a humidified 5% CO₂ environment. Mouse embryonic fibroblasts were grown to a 70 to 90% confluence before the next passage or for experiments. Cells were passaged by trypsinization, with 0.25% trypsin EDTA 1X (Thermo Fisher Scientific). Cells at passages 2 to 20 were used for the experiments.

To assess proliferation of mouse embryonic fibroblasts, cells were detached from the flask and centrifuged and rehydrated in the growing media. The cells were then seeded at a density of 2 × 10⁴ cells per well in a 12-well plate and incubated for 24 h, after which mouse embryonic fibroblasts were incubated with morphine sulfate (0.3 nM to 300 μM), fentanyl (0.3 pM to 300 nM), naloxone (3 μM), DMT-DALDA (0.3 nM to 30 μM), or PZM21 (0.03 μM - 30 μM) for 48 h at 37°C. The control group was incubated with the medium alone. The effect of drug on cell proliferation was determined by counting the number of cells in each well using hemocytometer by the observer blinded to the treatment.

In a limited assessment of the effects of morphine on fibroblast migration, mouse embryonic fibroblasts were plated onto the top of transwell plates (50 K · well^-1 · 200 µl^-1), precoated with collagen (1 µg/ml), and incubated for 4 h. Cells were counted after removing the top layer and staining. Cells were exposed to concentration of morphine, and then with naloxone at the optimal concentration. The control wells were incubated with the medium alone and wells to which fibroblast growth factor (13.5 ng/0.5 ml) was added served as the active comparator. Migration effects were presented as the % increase in the number of cells over control divided by the difference between the fibroblast growth factor well and control (e.g., the percent of the maximum effect on migration resulting from this concentration of fibroblast derived growth factor). Each series of concentration response curves were run at least three times and data pooled.

The mRNA levels of collagen type I α-2 were determined via quantitative polymerase chain reaction following the treatment with vehicle, morphine (10 μM), DMT-DALDA (10 μM), and PZM21 (10 μM). After the experimental treatment total RNA was isolated from the cells using TRIzol Reagent (Invitrogen, USA) and 1 μg of total RNA was used for cDNA synthesis by using iScript cDNA Synthesis Kit (Bio-Rad Laboratories, USA) according to the manufacturer’s instructions. cDNA was amplified using real-time polymerase chain reaction in an ABI 7300 real-time polymerase chain reaction system (Applied Biosystems, USA). RNA analysis reagents (SYBR Green Master Mix) were from Bio-Rad, USA. We used the comparative ΔΔ cycle threshold method to quantify collagen type I α-2 gene expression. Target gene expression levels in the test samples were normalized to the endogenous reference glyceraldehyde-3-phosphate dehydrogenase (F: 5´-CCA ACC GCG AGA AGA TGA CC-3´ and R: 5´-GAT CTT CAT GAG GTA GTC AGT-3´) levels and reported as the fold difference relative to glyceraldehyde-3-phosphate dehydrogenase gene expression in the untreated baseline control. All assays were performed in triplicate and the experiments were repeated at least three times.

Human embryonic kidney T cells stably expressing a β-arrestin-TEV fusion protein and a tTa-dependent luciferase reporter were maintained in Dulbecco’s Modified Eagle Medium (Corning, USA) containing 10% fetal bovine serum, 2 μg/ml puromycin, and 100 μg/ml hygromycin B at 37°C with 5% CO₂. Cells were plated to 50% confluency and transfected with a codon-optimized Mas-related G protein-coupled receptor-X Tango construct.²⁷  Transfected cells were transferred to glass-bottomed, poly-L-lysine-coated white 384-well plates (20,000 cells per well), and then were incubated for 18 to 24 h with concentrations of test article (in quadruplicate) diluted in drug buffer (1X Hank’s Balanced Salt Solution with 20 mM [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid] and 0.3% bovine serum albumin, pH 7.4). After incubation, medium was removed and 20 μl of Bright Glo (a luciferase assay; Promega, USA; diluted 20-fold) was added and incubated 15 min at room temperature. Luminescence was measured on a TriLux luminescence counter and analyzed as relative luminescent units using GraphPad Prism (GraphPad Software, Inc., USA). Additional details are provided elsewhere.²⁸ 

In the in vivo studies, no statistical power calculation to determine group size was conducted before the study. Group sizes were, however, projected to be a minimum of six guinea pigs per group, based on our previous work, showing that in the guinea pig model, the intrathecal concentrations of morphine employed in the current study produces increases in thermal escape latencies judged to be behaviorally relevant (6 to 8 s increases over baseline) and the ability to show significant increases in the incidence of intrathecal masses were statistically significant at the P < 0.01 (post hoc comparison with vehicle).⁸ 

Data were compiled in Excel (v.14.4.9, Microsoft Corporation, USA), and statistical analyses were performed using Prism (v.6.0).

Statistical analysis protocols for each data set were based on the results of testing the null hypothesis of normality (Kolmogorov–Smirnov test) and homogeneity of Bartlett test). Graphics of data sets for which the null hypotheses were not rejected present the data as means ± SD and employed parametric statistics (one-way or two-way repeated measures ANOVA analysis with post hoc comparisons being made using the Dunnett test). A summary of these assessments of normality and homogeneity of variance is presented in the table in Supplemental Digital Content (https://links.lww.com/ALN/B941). Data sets that were considered nonparametric (histology scoring) or for which treatment groups showed rejection of the normality hypothesis were analyzed using a one-way analysis using a nonparametric statistic (Kruskal–Wallace) and a post hoc comparison performed using a Dunn test. Differences in the post hoc analyses reaching a P < 0.05 of significance were considered to be significant.

A total of 71 guinea pigs were surgically prepared; 4 animals displayed motor dysfunction and did not recover and the catheter of one animal was lost after the start of infusion. These animals were euthanized and not included in subsequent analyses. A total of 66 animals were entered into and completed the studies. At necropsy, bolus injection of dye (10 µl: 0.5% weight by volume methylene blue) at study termination revealed uniformly intact catheters located within the intrathecal space at the level of the lumbar enlargement (approximately 11 cm from the cisterna magna).

Failure to reject the null hypothesis regarding normality for each treatment data set (Kolmogorov–Smirnov test) and homogeneity of variances (Bartlett test) for each set of treatments with the thermal escape latency/area under curve and mast cell degranulation data supported utilization of an ANOVA analysis. For the LAD2 studies, though groups sizes were too small to perform the Kolmogorov–Smirnov test, failure to reject the null hypothesis for homogeneity of variance and the observed covariance between means and medians led to use of ANOVA. For the fibroblast studies, where several treatment groups showed rejection of the normality hypothesis (see Supplemental Digital Content, table 1, https://links.lww.com/ALN/B941), we chose to employ a one-way repeated measures analysis using a nonparametric statistic.

The intrathecal infusion at 0.5 µl/h of saline (fig. 1A) had no effect, but morphine (33 nmol · 0.5 µl^-1 · h^-1 = 12.5 µg · 0.5 µl^-1· h^-1; fig. 1B), DMT-DALDA (10 pmol · 0.5 µl^-1 · h^-1 = 9.8 pg · 0.5 µl^-1 · h^-1), the potent μ agonist (fig. 1C), and PZM21 (28 nmol · 0.5 µl^-1 · h^-1 = 10.1 µg · 0.5 µl^-1 · h^-1, the biased μ ligand; fig. 1D) all resulted in a significant near maximal increase in the thermal escape latency by day 3. This increase displayed a loss of effect such that by day 7, there was an evident reduction in the peak thermal escape latency for all three agonists. Analysis of effects produced by continuous infusion was accomplished by comparing the area under the curve for each drug (fig. 1E). This analysis showed a significant main effect with the post hoc analysis revealing that all were statistically greater than intrathecal saline, but not different from each other.

To specifically determine the effects of opioid antagonism on the formation of the intrathecal mass, guinea pigs receiving intrathecal infusion of morphine were prepared to receive the concurrent continuous subcutaneous infusion of saline or naltrexone (25 µg/h; Supplemental Digital Content, fig. S1 and table S2, https://links.lww.com/ALN/B942). Animals receiving intrathecal morphine (33 nmol/h) plus subcutaneous saline for 14 days displayed a significant antinociception on days 3 and 7 (comparable to that observed in fig. 1, B and E) as compared to their respective baselines (day 0). With the exception of day 1, where a modest and transient, but statistically significant increase was noted, animals receiving subcutaneous naltrexone displayed a complete block of the analgesic action of the concurrently delivered morphine (Supplemental Digital Content, figure S1, A and B, https://links.lww.com/ALN/B942).

During the 14 to 16 days of infusion of saline, morphine, DMT-DALDA, or PZM21 at the indicated doses in animals shown in figure 1 and Supplemental Digital Content, figure S1 (https://links.lww.com/ALN/B942), there were no effects on pinnae or corneal reflexes. Motor functions (as defined by righting response, symmetrical ambulation, and hind paw proprioceptive function) were not altered in any animal at any time after the initiation of infusion, with the exception of one animal receiving intrathecal morphine on day 3 and one animal on day 1 receiving intrathecal DMT-DALDA, where mild hind limb motor weakness was noted. The only notable adverse events were the appearance of scratching and skin lesions in three animals receiving intrathecal morphine and a similar number in the group receiving intrathecal morphine and subcutaneous naltrexone. Scratching was largely absent after day 7 and skin lesions resolved. No such events were noted with DMT-DALDA or PZM21.

In the current work, we systematically assessed the histopathology of the spinal intrathecal space in catheterized guinea pigs after the 14 days of infusion of saline, morphine, DMT- DALDA or PZM21. In this work, the primary finding was the appearance of a space-occupying mass proximal to the lumbar catheter tip. These masses were graded on a 0 to 4 scale, as shown in the representative sections presented in figure 2, and separately by a forced ranking of the mass by size and degree of cord compression. Observer ranking concordance calculated as previously described¹⁰  showed that the intraclass correlation was found to be acceptable (intraclass correlation = 0.931), with an F value greater than 0.1, indicating no difference between raters, and a high level of concordance for the pathology grading. This analysis of the intrathecal mass by treatment group is presented in figure 3, A and B, respectively. As indicated, while the three opioids examined were delivered in approximately equianalgesic concentrations (fig. 1), only the intrathecal infusion of morphine resulted in prominent intrathecal masses. The median and interquartile range of the pathology grade for morphine (2.3: 2.0 to 3.5) were greater as compared to DMT-DALDA (1.0: 0.6 to 1.4), PZM21 (0.5: 0.5 to 1.8), and saline (1.3: 1.0 to 1.5).

An important question was whether the mass-producing effect of morphine was reversed by an ongoing opioid receptor antagonist. The continuous subcutaneous delivery of naltrexone, a pan opioid receptor antagonist, completely reversed the analgesic effects of intrathecal morphine (Supplemental Digital Content, figure S1, A and B, https://links.lww.com/ALN/B942), but on the contrary had no effect upon the intrathecal mass score or rank (fig. 3, A and B).

Histopathology emphasized several common observations. (1) As previously reported, the intrathecal mass in the guinea pig was localized to no more than 2 to 3 segments rostrally and caudally to the catheter tip, and was typically of sufficient girth as to yield varying degrees of local compression of the spinal cord. (2) Figure 4 demonstrates tissue from a guinea pig receiving intrathecal morphine showing the mass. Though closely adherent to the pial surface, it could be observed to be structurally distinct from the cord parenchyma with the demarcation being the pia (fig. 4C). Collagen staining tendrils could occasionally be observed to penetrate the pial layer. (3) In contrast, the mass was closely integrated with the adjacent dura-arachnoid. In this single section (fig. 4, B and D), the orientation and distribution of the fibroblasts can be seen apparently migrating from the dura into the intrathecal mass along with collagen-positive fibers. (4) The mass typically showed little inflammatory infiltrate and, based on the histopathology, was largely constituted of fibroblasts in a dense developing collagen matrix (fig. 4, B and C). (5) Herovici staining for collagen was predominately blue (emphasizing young, newly deposited collagen), with faint red staining (mature) in the outer regions of the mass, consistent with a mass growing from the inside (adjacent to the catheter and morphine infusate) (fig. 4E through G). Again, the Herovici stain emphasizes the demarcation of the mass at the pia (fig. 4F) and the collagen tendrils entering the mass from the dura-arachnoid (fig. 4G). (6) Using Picro-Sirius under bright field, dense collagen staining could be observed (fig. 4, H and I). With polarized light, the birefringence imaging revealed thick (yellow-orange birefringence) and to a lesser degree, thin (green birefringence) collagen fibers, indicative of types 1 and 3 collagen, respectively, could be readily observed (fig. 4J). As we have previously noted, the cellular composition of the mass in hematoxylin and eosin sections consisted of sparse macrophages and few lymphocytes.^8,9 

Our previous work with the morphine-evoked intrathecal mass emphasized the likely role of meningeal mast cells. In the current study, the absence of an effect upon mass formation of equianalgesic intrathecal infusion doses of DMT-DALDA and PZM21 raised the question of whether these agents could degranulate mast cells. Using the LAD2 human mast cell line, we confirmed that morphine indeed was a potent mast cell degranulator (fig. 5). In previous work we have shown in several models that the degranulating effects of morphine were not naloxone antagonized, but were prevented by the mast cell stabilizer, cromolyn.¹²  In contrast to morphine, PZM21 had no effect upon mast cell degranulation. DMT-DALDA resulted in a significant degranulation as compared to control, but only at the highest concentration (10 µM) examined.

To determine if morphine promotes mouse embryonic fibroblast proliferation, mouse embryonic fibroblasts were incubated with morphine at doses ranging from 0.3 nM to 300 μM. Morphine at 3 and 30 μM significantly enhanced fibroblast proliferation as compared to the control. A peak affect was observed at 3 μM (fig. 6A). Additional studies were carried out with the potent μ-opioid agonist (fentanyl; fig. 6B), the biased μ-opioid agonist PZM21 (fig. 6C), the potent μ-opioid agonist DMT-DALDA (fig. 6D). Though employed in a similar range of concentrations as used with morphine, these agents had no effect upon fibroblast proliferation.

To determine whether the proliferative effects of morphine were opioid receptor–mediated, fibroblasts were treated with a combination of morphine and naloxone in equimolar concentrations. As shown, naloxone did not inhibit morphine-induced proliferation of fibroblasts (fig. 6E). As cromolyn attenuated mast cell degranulation and prevented mass formation in the canine model,¹¹  we also examined the effects of cromolyn on the promotion by morphine of fibroblast proliferation. Here we found that cromolyn indeed reversed the proliferation otherwise produced by morphine in the absence of mast cells (fig. 6F).

Morphine induced significant mouse embryonic fibroblast migration at doses 2 to 20 µM, respectively. Naloxone at equimolar and up to 10× the equimolar concentrations of morphine (20 µM) did not prevent migration induced by morphine (Supplemental Digital Content, figure S2 (https://links.lww.com/ALN/B943).

Treatment of mouse embryonic fibroblasts with morphine (10 µM) within the concentration range producing fibroblast proliferation resulted in a significant increase in collagen type I α-2 mRNA as measured by quantitative polymerase chain reaction. In contrast, treatment with equimolar concentrations of DMT-DALDA (10 μM) or PZM21 (10 μM) had no effect as compared to buffer control (fig. 7).

Current work has shown that an important cascade for degranulating mast cells is through the activation of Mas-related G protein-coupled receptor-X2. Using the PRESTO-Tango assay, we assessed the effects of these several opioid ligands on their ability to activate Mas-related genes. As shown in figure 8, Tan 67 (a δ-opioid agonist), morphine and DMT-DALDA all resulted in near maximal degrees of activation with the Log EC50 (M) being −5.8, −5.1, and −4.7, respectively. In contrast, as fentanyl and the biased μ ligands PZM21 and TRV130 failed to produce a saturating curve, the EC50s could not be calculated and the compounds can be said only to show some activity in the PRESTO-Tango assay at approximately 100 µM. These results with morphine, DMT-DALDA, and PZM21 are consistent with the relative effects of these agents on mast cell degranulation (fig. 5) and fibroblast activation (fig. 6 and 7). Given the ability of cromolyn to prevent mast cell degranulation and fibroblast activation, separate studies were undertaken to determine if cromolyn activated Mas-related genes in the PRESTO-Tango assay. It did not. Further, coincubation of cromolyn with morphine or a full agonist (mitiglinide) or a partial agonist at the Mas-related G protein-coupled receptor-X receptor did not alter their evoked Mas-related gene activation (Supplemental Digital Content, figure S3, https://links.lww.com/ALN/B944). These results indicate that cromolyn is neither an agonist nor an antagonist at the Mas-related G protein-coupled receptor-X receptor, as measured by the Tango arrestin recruitment assay.

Intrathecal opiate use is impacted by the risk of producing space-occupying intrathecal masses. The present work provides specific insights into mechanisms whereby opiate agonists initiate intrathecal masses by signaling through the noncanonical Mas-related gene pathway.

The first issue is that the agents examined at the respective doses resulted in an equianalgesic activity. Morphine, PZM21, a biased μ ligand, or DMT-DALDA, a dermorphin-derived μ-opioid peptide, were delivered by intrathecal infusion in concentrations producing a comparable increase in escape latencies. The molar potency ratio of these molecules in the infusion paradigm was approximately 1: 0.8: 3000. All agents resulted in analgesia by days 2 to 3 and all showed loss of effect by day 7. This timeframe is observed in the rat and guinea pig.^29,30  While the role of mass formation in loss of analgesia for morphine was not assessed, this loss of effect was observed with PZM21 and DMT-DALDA, which did not produce a mass. Although some suggest that biased ligands may show differences in development of tolerance,^31–33  this was not observed after intrathecal (current studies) or systemic delivery.³⁴  At these doses, minimal adverse effects were observed with either PZM21 or DMT-DALDA. Morphine was characterized by an incidence of caudally directed scratching and skin lesions during the first 7 to 10 days, which resolved.

Intrathecal masses arise from the dura-arachnoid proximal to the site of intrathecal delivery (fig. 4).^6,9  While referred to as an “inflammatory” mass, it is largely composed of fibroblasts in a dense collagen matrix.^8,9  Intrathecal fibrosis is observed after a variety of pathologies (e.g., subarachnoid hemorrhage) which, in humans, as in the intrathecal mass observed in guinea pig and dog, is composed of collagen types 1 and 3.^8,35,36  The mass presents as a progressive expansion from the catheter outwards with less mature (newly created) collagen observed proximal to the catheter and mature collagen present in the outer margins.⁸  Sequential magnetic resonance imaging in dogs receiving morphine infusion revealed a time-dependent enlargement of the mass.^6,37  In early stages, substitution of saline results in a reduction in mass size. Termination in the late phase reveals a mature mass that resolves slowly, consistent with the mature collagen outlining the masses in guinea pig and dog.⁸ 

While a catheter can initiate a reaction,⁸  the enlarging mass is not a simple response to catheter or infusion,^6,38  infection,^6,39  or formulation.¹  Preclinical studies show that intrathecal masses are produced by several (morphine, hydromorphone, methadone), but not all (fentanyl, alfentanil) opioids,^6,10,11  a profile largely recapitulated in humans.^40,41  Further, μ-opioids DMT-DALDA^18,19  and PZM21⁴²  resulted in analgesia but failed to produce an intrathecal mass. This profile is consistent with failure of opioid antagonism to block formation of masses in dogs¹  and, as shown here, in guinea pigs.

In primary mast cell cultures and in the wheal-flare response after subcutaneous delivery in dog, three observations were made. (1) Agents that degranulate mast cells (morphine, hydromorphone, and methadone) result in an intrathecal mass, while those that do not degranulate (fentanyl or alfentanil) have not been reliably associated with spinal masses.^10–12  (2) Degranulating effects of morphine were not blocked by naloxone¹²  and neither were the effects of intrathecal morphine in producing masses (current studies).⁹  (3) Further, in dogs, cromolyn sodium (mast cell stabilizer) diminished degranulation/fibroblast activation and attenuated development of morphine-induced spinal masses.⁹  In the current work, we confirmed this association by showing that PZM21 failed to produce degranulation/activation or a mass while DMT-DALDA did so in concentrations comparable to those of morphine, but the equianalgesic intrathecal concentration for DMT-DALDA was in the pmol/h versus nmol/h range.

Mas-related G protein-coupled receptors are classified as orphan G protein-coupled receptors and implicated in IgE-independent inflammatory responses in mast cells.^14,43  Morphine-evoked increases in intracellular calcium and mast cell degranulation occur through Mas-related G protein-coupled receptor signaling.¹⁵  Using the PRESTO-Tango assay, morphine and DMT-DALDA resulted in activation of Mas-related G protein-coupled receptor-X2. In contrast, PZM21 and fentanyl produced a degree of activation only at 100 µM. While saturating concentrations for these drugs in this assay was not examined, these results correspond to the modest effect of these agents on mast cell degranulation/fibroblast activation where, in contrast, morphine yielded the greatest effects. DMT-DALDA, a μ-opioid peptide, was only modestly active and then only so at the highest concentration. PZM21 had no effect on mast cell degranulation. Though not studied here, our previous work with fentanyl in human mast cell cultures and dog skin after intradermal injection displayed, in contrast to morphine and hydromorphone, little or no effect on mast cell degranulation.¹² 

Intrathecal infusion at equianalgesic concentrations of morphine, but neither PZM21 nor DMT-DALDA, yielded an intrathecal mass. The lack of effect of PZM21 corresponds with its lack of effect on Mas-related G protein-coupled receptors and on its failure to yield mast cell degranulation and fibroblast activation. While DMT-DALDA yielded measurable degranulation and activation at concentrations comparable to those employed by morphine, the DMT-DALDA concentration required was 0.001× that required for morphine. Accordingly, failure to see a mass with DMT-DALDA indicates that concentrations necessary to produce analgesia were substantially less than those concentrations required to produce meningeal mast cell degranulation. A similar circumstance holds for the Cav 2.2 channel blocker, ziconotide, which is a potent mast cell degranulator,¹²  but is not considered to be associated with mass formation. This reflects the low intrathecal ziconotide concentrations required for analgesia versus concentrations required to produce mast cell degranulation.¹²  Based on the above, we predict that TRV130 (oliceridine), a biased μ ligand that did not activate Mas-related genes and did not degranulate mast cells or activate fibroblasts, would not result in a mass. In contrast, TAN67 (a δ-opioid agonist) would yield all three effects, if, unlike DMT-DALDA, it required a high dose to produce analgesia.

The intrathecal mass is a fibroblast-rich collagen matrix arising from the meninges. In fibroblast cultures, absent mast cells, morphine, in a naloxone-insensitive fashion, but not PZM21 or DMT-DALDA in concentrations comparable to those studied in the degranulation of mast cells, resulted in fibroblast proliferation in cell culture which was prevented by cromolyn. Quantitative polymerase chain reaction for collagen type I revealed that morphine, but not DMT-DALDA or PZM21, increased collagen type 1 mRNA, suggesting that fibroblast activation may also be accomplished in an Mas-related gene–dependent fashion. Microarray data bases indeed report that human cutaneous fibroblasts involved in keloid scar formation express the Mas-related gene family, with the rank ordering of expression being Mas-related genes 2 and 4 greater than 1 and 3.⁴⁴ 

Our hypothesis associating Mas-related gene signaling in mast cells and fibroblasts and mass formation hinges on four observations derived from cultured mast cells and fibroblasts and in vivo spinal infusions: (1) an opiate that activates Mas-related genes, degranulates mast cells and activates fibroblasts produces a spinal mass (morphine); (2) an opiate that does not activate Mas-related genes, does not degranulate mast cells, or activate fibroblasts does not produce a spinal mass (PZM21); (3) an opiate which activates Mas-related genes, degranulates mast cells, and activates fibroblasts, but is significantly more potent as a μ agonist does not produce masses at equianalgesic concentrations to morphine; and (4) a mast cell stabilizer (cromolyn) blocks morphine induced mast cell degranulation, fibroblast activation, and as previously reported, reduces mass formation. The cellular mechanism of cromolyn is not well defined.⁴⁵  We exclude a direct interaction with Mas-related genes. The present work suggest that the fibroblast rich mass results from at least two nonexclusive processes.

• (1) Indirect pathway to fibroblast activation. Degranulation of meningeal mast cells through Mas-related gene engagement leads to release of tryptases and chymases, inducing fibroblast proliferation and migration,^13,46–48  while blocking degranulation reduces dermal fibroblast chemotaxis and inhibiting mast cell function decreases scar formation.^46,47,49,50 

• (2) Direct pathway to fibroblast activation. Activation of Mas-related genes in meningeal fibroblasts lead to proliferation, migration and collagen secretion. This mechanistic profile closely resembles events underlying keloid scarring where exuberant fibroblast activation is enhanced by mast cell activation. Hence, we may consider the intrathecal mass to be a keloid scar of the meninges.

• (3) An additional pathway may also exist. A spinal mass resected from a patient receiving 50 mg/ml displayed morphine particulates.⁵¹  Intrathecal particulates (methylprednisolone acetate,⁵²  talcum powder, and starch⁵³ ) can also initiate meningeal reactions.

The direct effect of opioids on fibroblasts and mast cells are demonstrated in in vitro cell culture experiments. The linkage between in vitro studies showing activation and spinal masses (in vivo) is indirect, based on parallels between the Mas-related gene agonist pharmacology and inhibitory effects of cromolyn on mast cell and fibroblast activation and spinal mass formation by agents activating Mas-related genes.

Incidence of intradural masses with chronic opiate infusion is estimated to be around 8%.⁴⁰  The role of concentration presents a conundrum for intrathecal delivery given the need to provide sufficient activation with infusate volume constraints posed by implanted pumps. Agents that do not activate Mas-related genes (PZM21) or do so at concentrations above those required for analgesia (DMT-DALDA) may provide an alternate approach to activating potent regulatory control over spinal nociceptive processing.

The authors would like to thank Brian Shoichet, Ph.D., Department of Pharmaceutical Chemistry, University of California, San Francisco, California, and Bryan Roth, M.D., Ph.D., Department of Pharmacology, University of North Carolina, Chapel Hill, North Carolina, for their specific commentaries on the interpretation of the PRESTO-TANGO assay results and the properties of biased opiate receptor ligands, and Julio Medina, Ph.D., of Epiodyne, San Francisco, California, for providing the PZM21. The support for the Tango assays came from the National Institute of Mental Health Psychoactive Drug Screening Program (Bethesda, Maryland; Dr. Roth). The authors also thank Matt Hunt, B.S., M.S., University of California, San Diego, California, for his important input on Mas-related gene expression in fibroblasts. The authors would also like to express their recognition of the role played by Frederick W. L. Kerr, M.D. (1923 to 1983), Department of Neurosurgery, Mayo Clinic, Rochester, Minnesota, in advancing the therapeutic development of spinal opiates, Burton M. Onofrio, M.D., Department of Neurosurgery, Mayo Clinic, for his implementation of the first intrathecal pump, and Christopher M. Bernards, M.D. (1958 to 2012), Virginia Mason, Seattle, Washington, who suggested the importance of mast cells in the intrathecal mass.

Early portions of the in-life guinea pig work examining naltrexone effects of the morphine mass was supported by funds from Medtronic (Minneapolis, Minnesota) provided to Dr. Yaksh. Dr. Hildebrand and Dr. Page are Medtronic employees and contributed to the design of the studies and preparation of the final manuscript. Dr. Griener holds a patent position on the discovery of PZM21. Dr. Schiller holds a patent position on the discovery of DMT-DALDA.

This work was supported by National Institutes of Health grant No. R01 NIDA015353 (Bethesda, Maryland; to Dr. Yaksh) and funds from Medtronic (Minneapolis, Minnesota) supporting early portions of the in vivo guinea pig work.