Monosodium urate monohydrate crystal–induced inflammation in vivo: Quantitative histomorphometric analysis of cellular events




To quantify the inflammatory cell response in rat air pouch pseudosynovial membrane during monosodium urate monohydrate (MSU) crystal–induced inflammation.


In the rat air-pouch model, we used a computer-assisted histomorphometric method to quantify cell distributions, based on cell linear densities, in histologic sections of membranes from pouches injected with MSU or saline. The volume, white blood cell (WBC) count, and histamine content of the pouch exudates were determined at several time points.


Injection of 10 mg of MSU crystals into the pouch produced an acute exudate. After peaking at 24 hours, the exudate volume and WBC count decreased spontaneously over the next 3 days, simulating the self-limited course of acute gout. Membrane thickness followed a parallel course. Membrane polymorphonuclear cell (PMN) linear densities were closely correlated with exudate WBC counts, suggesting PMN recruitment from the subintimal synovial membrane. Both monocyte/macrophage and mast cell linear densities increased in the subintimal layer 2 hours after crystal injection (P = 0.038 and P = 0.03, respectively, versus controls), whereas PMN linear densities showed 2 peaks, one at 4 hours and the other 24 hours. The exudate histamine content peaked 6 hours after crystal injection, when mast cell linear densities were minimal in the membranes, suggesting mast cell degranulation.


An increase in monocyte/macrophage and mast cell densities in the membrane preceded the PMN influx in the pouch membrane and exudate, suggesting that mast cells may be involved in the early phase of MSU crystal–induced inflammation, at least in this rat model.

The pathogenesis of acute gout has been addressed in many studies. Synthetic monosodium urate monohydrate (MSU) crystals have been injected into the tissues or joints of various animals, including dogs (1), rats (2), mice (3), and rabbits (4). These in vivo experiments have elucidated the kinetics of acute-phase proteins, cytokines, and polymorphonuclear cell (PMN) and mononuclear cell (MNC) contents in exudates, showing a peak in cell recruitment and demonstrating that the acute phase was followed by self-limitation of the MSU crystal–induced inflammation. In vitro studies have shown that MSU crystals added to various cell cultures (5) induced time- and dose-dependent cytokine secretion. Schumacher et al (1) and Gordon et al (6) have reported evidence that acute gout is initiated by phagocytosis of free MSU crystals by the lining cells (type A [macrophage-like] synovial cells). This event is currently acknowledged to precede synovial inflammation per se (7, 8). However, little is known about the kinetics of the cellular events within the synovial membrane.

In addition to the initial role played by macrophage-like lining cells, cell activation is believed to trigger massive recruitment not only of PMNs, which are recognized as the main inflammatory cells in acute gout, but also of mast cells, MNCs, and endothelial cells associated with acute neoangiogenesis. Few qualitative pathologic studies of human or animal synovial tissues have been reported (9, 10), and we are not aware of systematic quantitative studies of cellular changes over time. Therefore, in the rat air-pouch model of the synovial cavity, we used a quantitative histomorphometric method (11) to evaluate the inflammatory response induced in the pouch pseudomembrane by MSU crystal injection.


Crystal preparation. Synthetic MSU crystals were prepared as described previously (12). We added 1.68 gm of uric acid (ICN Biomedicals, Aurora, OH) to a 0.01M NaOH solution heated to 70°C. NaOH/HCl was added as required to keep the pH between 7.1 and 7.2. The solution was stirred slowly and continuously at room temperature. Twenty-four hours later, the crystals were harvested by decanting the supernatant; they were then washed, dried, dispensed into individual vials (10 mg), and sterilized by autoclaving. The needle shape and size of the crystals were checked by polarizing light microscopy. The absence of endotoxin contamination was verified using a Limulus amebocyte cell lysate assay (E-toxate kit; Sigma, St. Louis, MO). The sterile MSU crystals were resuspended in 5 ml of sterile saline solution just before injection into 6-day-old rat air pouches.

Rat air-pouch model. We used the rat air-pouch model of the synovial cavity (13). All experiments were conducted according to European ethical guidelines. Noninbred Sprague-Dawley rats weighing 130–150 gm at the time of the experiment (Centre d'Elevage R Janvier, Le Genest-St. Isle, France) were used. Five rats were housed per cage, fed normal chow, and maintained in accordance with current standards for the confinement of laboratory animals. The experiments were performed with the rats under anesthesia induced by intraperitoneal ketamine injection.

Twenty milliliters of sterile air was injected subcutaneously through a 0.25-μm microfilter into the backs of the animals to create a pseudosynovial cavity. A second air injection was given on day 3 to keep the pouch inflated. Six days after the first air injection, 5 ml of sterile saline solution (control group) or 10 mg of MSU crystals resuspended in sterile saline solution (MSU group) was injected into the pouch. Groups of 6–8 the animals were killed by cervical dislocation under anesthesia at 1, 2, 4, 6, 24, and 48 hours after the crystal or saline injections.

Collection and processing of samples. The exudates and pseudosynovial membranes from the air pouches were harvested. Exudates were collected using a catheter under sterile conditions and were immediately cooled on ice before processing. Exudate volumes were measured, and white blood cell (WBC) counts were obtained using a standard hemocytometer. Differential WBC counts were determined after cytospin centrifugation followed by May-Grünwald–Giemsa staining. Histamine contents were measured using the radioenzymatic assay described by Haimart et al (14).

Pseudomembranes were dissected, fixed in 10% buffered formalin (pH 7.2–7.4), dehydrated in ethanol, and embedded in Paraplast. Random 5-μm sections (20 per pouch) were cut using a Minot-type microtome and routinely stained with Masson's trichrome or hematoxylin–eosin–saffron (HES). These sections were used for the quantitative light microscopy analysis. Other sections were stained with toluidine blue (pH 4.2) and used to identify and count mast cells (15).

For transmission electron microscopy (TEM), small pieces of membrane were fixed in 2.5% glutaraldehyde solution, postfixed in 1% osmic acid solution, and, after gradual dehydration in alcohol, embedded in SPURR resin (TABB, Berkshire, UK). Ultrafine sections ∼90-nm thick were examined under an EM300 microscope (Philips, Eindhoven, The Netherlands).

Computer-assisted cell counting. To quantify changes in pouch membranes, we used a method previously developed by Christel and Meunier for determining cell distributions in tissue encapsulating surgically implanted biomaterials (11). Sections were observed at 400× or 1,000× magnification under a BHT microscope (Olympus France, Rungis, France) equipped with a drawing tube. The image of the histologic section was digitized using the diode cursor of a graphic tablet connected to a microcomputer, with the diode superimposed on the histologic slide through the drawing tube. For each optical field, membrane contours and vessels were digitized at 400× magnification. The nuclei of each cell type (fibroblasts, macrophages, PMNs, and mast cells) and the blood vessels were digitized at 1,000× magnification. During the digitization process, the computer recorded the coordinates of the cell nuclei with reference to the membrane's inner contour (Figure 1). The procedure was repeated on at least 5 separate fields of the membrane, and the data from these 5 fields were pooled.

Figure 1.

Parameters evaluated by histomorphometry. Membrane contours, vessels, and each cell type were digitized using a histomorphometric apparatus with 400× or 1,000× magnification. Membrane thickness and linear densities (LDs) of vessels and of each cell type (fibroblasts, macrophages, polymorphonuclear cells, and mast cells) were computed per millimeter of digitized membrane interface. Because LDs were not normally distributed (see graph in box), the nonparametric Mann-Whitney test was used for statistical analysis. d = mean length of the membrane border.

The following parameters were calculated: 1) distance parameters, including membrane thickness as well as the median and quartile (25%, 75%) distances of each cell type from the pouch lining; 2) linear densities, taking into account the number of cells per millimeter of digitized membrane; and 3) cumulative counts of each cell type in 5-μm sections with no gap, from the lining to the subintimal layer. Because no significant cell changes occurred between 0 μm and 20 μm from the surface, this area was arbitrarily defined as the lining, and the area from 20 μm to 80 μm, where the tissue was looser, was defined as the subintima.

Statistical analysis. For normally distributed parameters, namely, membrane thickness, vessel numbers, and WBC counts, as well as histamine concentrations, results were expressed as the mean and SD. Between-group comparisons were performed using one-way analysis of variance, and differences between means were determined using Student's t-test.

Linear densities of fibroblasts, macrophages, PMNs, and mast cells were not normally distributed. For these parameters, results were calculated as the median ± quartiles and were compared using the nonparametric Mann-Whitney test. Linear correlations were determined using the nonparametric Spearman's rank correlation test.

Validation of the quantitative and histomorphometric method.Intraobserver and interobserver errors. Intraobserver error was calculated by digitizing the same histologic section 5 times. Intraobserver error was 0.9% for membrane thickness, 5% for fibroblast densities, and 12% for macrophages. Interobserver differences determined by having two observers digitize the same 5 sections were not statistically significant for any cell types (data not shown).

Staining and magnification. Cell counts were not significantly different with Masson's trichrome or HES staining. However, optical magnification significantly influenced the cell counts (P < 0.01), with the most reliable results being achieved at high magnification (1,000×). The various parameters were representative of the slide only if at least 5 fields on the slide were digitized; digitizing more than 5 fields did not modify the findings.

Similarly, serial sections of air-pouch membrane were cut, digitized, and examined. No significant differences were found across the slides (data not shown). Consequently, for the study, a single slide taken at random was digitized for each air pouch.


Qualitative analysis of the pseudosynovial membrane not exposed to MSU crystals. The pouch membrane comprised 3 layers, as previously described (16). The lining layer or intima, was well defined, with 1–3 layers of cells, including flattened fibroblasts and macrophages. Rare PMNs were scattered among these two cell types. The subintimal layer was a looser tissue with an abundance of cells, including fibroblasts, macrophages, PMNs, lymphocytes, and mast cells, as well as many small vessels and capillaries. A few MNC aggregates were found around the vessels. Under the subintimal layer, there was an areolar zone, with fewer cells and a greater number of blood vessels.

Quantitative histomorphometry. Membrane thickness increased transiently after MSU crystal injection. Compared with control membranes, the increase was significant 6 hours after the injection (P = 0.004) (Figure 2). Maximal thickness was achieved after 24 hours (P = 0.001 and P = 0.004 versus thickness after 6 hours and 48 hours, respectively). No difference in vessel number, as determined by counting vessel and capillary sections, was found between injected and control pouches (data not shown).

Figure 2.

Kinetics of air-pouch membrane thickness. Values are the mean and SD. Comparison between the control group and the monosodium urate monohydrate (MSU) crystal–induced inflammation group was performed by Student's t-test.

Kinetics of fibroblast, monocyte/macrophage, PMN, and mast cell linear densities.Fibroblasts. Only one significant difference was observed between the two groups. This consisted of a significant decrease in median fibroblast linear densities 4 hours after crystal injection (P = 0.03) (data not shown).

Monocyte/macrophages. Compared with control tissues, MSU crystal injection resulted in a significant increase in monocyte/macrophage linear densities as early as 2 hours after crystal injection (P = 0.038) (Figure 3A). Between 1 hour and 2 hours after crystal injection, median monocyte/macrophage linear density increased by 160%, from 48 cells/mm to 125 cells/mm, in the MSU-injected animals. The monocyte/macrophage linear density increase over time was not significant in the lining (the most superficial 20 μm of depth), but was highly significant in the subintima (20–80 μm from the surface) in both the MSU and control groups (Figure 4A). Wide variations in monocyte/macrophage linear densities over time were noted in the subintima of the MSU-injected membranes, whereas the increase was steadier in the control membranes. Importantly, monocyte/macrophage linear densities remained high in the subintima after 48 hours (110 cells/mm), although inflammation in the pouch exudate had diminished by then.

Figure 3.

Kinetics of monocyte/macrophage, polymorphonuclear cell (PMN), and mast cell linear densities in the air-pouch membranes after injection of monosodium urate monohydrate (MSU) crystals. Values are the median. A, A significant increase in monocyte/macrophage linear density was found after 2 hours in the MSU group (P = 0.038 versus control group). B, Two peaks in PMN linear densities were seen at 4 hours and 24 hours after crystal injection (P = 0.04 and P = 0.05, respectively, versus control group). PMN linear density remained <5 cells/mm in the control group. C, Mast cell linear densities peaked 2 hours after crystal injection (P = 0.03 versus control group). No mast cells were detected by toluidine blue staining 6 hours after crystal injection. A low mast cell linear density peak was observed after 6 hours in the control membranes.

Figure 4.

Kinetics of monocyte/macrophage, polymorphonuclear cell (PMN), and mast cell counts within the membrane. Cumulative numbers of A, monocyte/macrophages, B, PMNs, and C, mast cells were determined in consecutive 5-mm segments from the lining to the subintimal layer. The 20-μm line was used as a cutoff in all cases because significant cell changes occurred only below this line (20–80 μm from the surface). MSU = monosodium urate monohydrate.

PMNs. In controls, median PMN linear densities showed little change over time, remaining at very low values (<5/mm after 24 hours). In the MSU crystal–injected animals, 7–13 PMNs/mm were counted within the membrane as early as 4 hours after crystal injection (Figure 3B). A second PMN density peak (25/mm) was seen 24 hours after crystal injection. Compared with saline, MSU crystals significantly increased PMN linear densities at 4 hours and 24 hours (P = 0.04 and P = 0.05, respectively). PMN densities returned to values similar to those in the controls at 48 hours. As with the monocyte/macrophages, this transient increase in the number of PMNs was seen in the subintima, but not in the lining (Figure 4B). Median PMN linear densities were strongly correlated with contemporaneous exudate WBC counts (r2 = 0.91, P = 2 × 10−6) (data not shown), which decreased spontaneously after peaking at 24 hours.

Mast cells. Again, mast cells identified by toluidine blue staining (Figure 5A) and characteristic ultrastructure (Figure 5B) were found only in the subintima (Figure 4C). Mast cell linear density peaked as early as 1–2 hours after crystal injection, whereas no mast cells were detected in control membranes at this early time point (P = 0.03) (Figure 3C). Inflammatory membranes contained 17–24 mast cells/mm within 2 hours after crystal injection (Figure 3C). In contrast, 6 hours after crystal injection, no mast cells were detected by toluidine blue staining. The histamine content in pouch exudates peaked 6 hours after crystal injection, when mast cell linear densities assessed by toluidine blue staining were minimal in the membrane (Figure 6), a finding consistent with early degranulation. In control membranes, a significant, but lower, mast cell linear density peak was achieved at 6 hours (P = 0.002) (Figure 3C).

Figure 5.

Micrographs of a mast cell in a membrane. A, Toluidine blue staining of a mast cell. B, Electron micrograph of a mast cell in a membrane, sampled 2 hours after monosodium urate monohydrate crystal injection. (Original magnification × 1,000 in A; × 11,700 in B.)

Figure 6.

Kinetics of histamine levels in air-pouch exudates associated with mast cell linear density in the membranes. Mast cell linear densities were expressed as medians. Histamine levels were measured in supernatants. The histamine content in pouch exudates peaked 6 hours after monosodium urate monohydrate crystal injection, when mast cell linear densities assessed by toluidine blue staining were minimal in the membranes, suggesting an early degranulation.

As can be seen in Figure 3, comparison of mast cell and PMN linear density kinetics in the crystal-injected pouches showed that the PMN peak was preceded by an early mast cell peak. A similar early monocyte/macrophage increase was observed 2 hours after crystal injection.


We used a quantitative histomorphometric method to determine the kinetics of cellular changes in an experimental rat air-pouch model of acute MSU crystal–induced inflammation. This acute inflammation was characterized by an early and transient increase in leukocyte linear densities within the membrane subintimal layer, the kinetics of which correlated with the PMN influx (r2 = 0.91, P = 2 × 10−6) in pouch exudates. Macrophage linear density increased significantly after 2 hours in the sublining around the vascular areas, and macrophages were the most abundant inflammatory cells in the membrane, a finding consistent with descriptions of the synovium in human gout (10). Interestingly, we also observed mast cell infiltrates in the subintimal layer; these developed at the same time as the monocyte/macrophages and before the PMN recruitment to the membrane. Determinations of histamine levels in pouch exudates, specific toluidine blue staining, and TEM of the pseudosynovial membrane confirmed that early mast cell degranulation occurred. The increase in cell linear density was observed only in the subintima, at levels deeper than 20 μm from the surface. Inflammatory cell recruitment from the bloodstream may occur in the subintima, since this area contains constitutive vessels. The absence of linear density changes in the lining (within 20 μm of the surface) suggests cell migration from the membrane through the cavity.

The histomorphometric technique we used was developed to evaluate the biocompatibility of implants in rodents (11). This time-consuming technique allowed us to perform a comprehensive statistical analysis of several parameters reflecting the distribution of various cell populations over time. Thus, we were able to obtain a detailed picture of normal and inflammatory pseudomembranes. We did not take into account the functional status of cells, which could be studied, for example, by immunochemistry and in situ hybridization for pro- and antiinflammatory cytokines.

Membrane thickness was significantly increased 6 hours after MSU injection compared with the membrane in control pouches. This rapid increase in thickness probably reflects an early increase in vascular permeability, with egress of protein and water resulting in membrane edema. The increase was contemporaneous with mastocyte degranulation, which releases various stored materials, among which only histamine was measured in this study. This bioamine release might have contributed to the observed variation in membrane thickness through various transduction couplings (e.g., Na+/H+ exchange) (17).

There was no difference in the number of capillaries or venules between control and MSU-injected membranes, although inflammation is believed to be associated with neoangiogenesis. This intriguing observation may be ascribable to the fact that we investigated neither vessel types nor endothelial cell activation. We examined only the number of vessels, irrespective of their diameter, type (arterioles, venules, or capillaries), and location in or outside of the lining. A study in an animal model of acute crystal-induced inflammation has shown that intracutaneous MSU crystal injection activated capillary endothelial cells, as demonstrated by E-selectin expression (18). Interestingly, colchicine down-regulates E-selectin expression in vitro (19) and in vivo (20), leading to reduced leukocyte migration in inflamed tissues. The antiinflammatory effect of antiselectin antibodies in a model of acute peritoneal MSU crystal inflammation in rodents further supports a role for endothelial cell activation through selectin expression (21).

One interesting aspect of our experiment is that it allowed us to compare changes over time in the numbers of various inflammatory cell types in the membrane. Importantly, the monocyte/macrophage peak occurred 2 hours after crystal injection, preceding the PMN peaks, which occurred 4 hours and 24 hours after crystal injection. Monocyte/macrophages are known to release preformed and de novo–synthesized chemotaxins and endothelial cell–activating factor (22) and may therefore play a major role in PMN recruitment to the membrane and pouch exudate. However, one limitation of our study is that we did not determine the activation status of membrane monocyte/macrophages; this needs to be done to test the hypothesis of a role for monocyte/macrophages.

As demonstrated for monocyte/macrophages, changes in PMN numbers were more striking around the vessels in the subintimal layer after crystal injection. Together with the highly significant correlation between PMN linear densities in the membrane and WBC counts in the exudate, this suggests that PMNs were recruited from the subintimal layer at the early phase and that they migrated into the pouch cavity during the acute phase. PMNs, which are the predominant cells found in the synovial fluids of patients with acute gout, are believed to play a pivotal role in MSU-induced inflammation (23). Their ability to phagocytose crystals, as well as to release numerous mediators, including chemotactic factors, indicates that PMNs contribute to amplifying MSU crystal–induced inflammation.

Interestingly, we also found an early increase in the number of mast cells in the subintimal layer of MSU-injected pouch membrane, suggesting that recruitment of mast cells from the blood occurred during the early phase of MSU crystal–induced inflammation. The increase in mast cell linear density started before the monocyte/macrophage and PMN peaks, supporting the hypothesis that mast cells play a role in the early phase of MSU-induced inflammation, which involves vascular exudation and cell attraction.

Mast cells are present in small numbers in the normal synovial membrane and in larger numbers in the rheumatoid arthritis pannus (24–26). They have not been observed in human synovium from joints with acute gout, although specimens have not been obtained during the very early stages of synovial inflammation. Mastocytes are known to generate and to release several biologically active molecules following activation by IgE-mediated stimuli, complement-derived anaphylatoxins, lymphokines, and monokines (27). The numerous compounds released by mast cells include vasoactive amines, which are potent exudation factors, and chemotactic factors that promote PMN influx, such as the chemokine epithelial neutrophil activating peptide 78 (22). Getting et al (21) have reported that mast cell mediators may contribute to the PMN influx generated by peritoneal MSU crystal injection in mice. In this model, a role for endogenous histamine and platelet-activating factor (PAF) was supported by the inhibitory effect of PAF inhibitor and of an anti-H1 histamine receptor antagonist.

Taken together, these results demonstrate that early infiltration of the pseudosynovial membrane by monocyte/macrophages and activated mast cells precedes the PMN influx triggered by MSU crystal injection. Thus, synovial mast cells may be involved in the early phase of gouty inflammation. Our study shows that both exudates and membranes should be analyzed in parallel to gain insight into cell kinetics during experimental gouty inflammation.