Treatment of Murine Colitis by Lactococcus lactis Secreting Interleukin-10

See allHide authors and affiliations

Science  25 Aug 2000:
Vol. 289, Issue 5483, pp. 1352-1355
DOI: 10.1126/science.289.5483.1352


The cytokine interleukin-10 (IL-10) has shown promise in clinical trials for treatment of inflammatory bowel disease (IBD). Using two mouse models, we show that the therapeutic dose of IL-10 can be reduced by localized delivery of a bacterium genetically engineered to secrete the cytokine. Intragastric administration of IL-10–secretingLactococcus lactis caused a 50% reduction in colitis in mice treated with dextran sulfate sodium and prevented the onset of colitis in IL-10−/− mice. This approach may lead to better methods for cost-effective and long-term management of IBD in humans.

Inflammatory bowel disease (IBD), including Crohn's disease and ulcerative colitis, is a significant public health problem in Western societies, affecting 1 in 1000 individuals, yet its etiology remains poorly understood. IBD is characterized clinically by chronic inflammation in the large and/or small intestine, the symptoms of which include diarrhea, abdominal pain, weight loss, and nausea. Death can result, in extreme cases, from malnutrition, dehydration, and anemia. IBD is thought to arise from interacting genetic and environmental factors (1) and may involve abnormal T cell responses to commensal microflora (2–4). Biologically based therapies such as antibodies to tumor necrosis factor (TNF), which is a strong proinflammatory mediator (5–7), and recombinant IL-10 (8) can ameliorate the disorder.

Because IL-10 has a central role in down-regulating inflammatory cascades (9) and matrix metalloproteinases (10), it is a likely candidate for use in therapeutic intervention. In this study we have tested a new method of delivering IL-10: in situ synthesis by genetically engineered bacteria (Lactococcus lactis), in two mouse models of the disease, one involving treatment of chronic colitis induced by 5% dextran sulfate sodium (DSS) (11) and one involving prevention of colitis that spontaneously develops in IL-10−/−mice (12). We show that this approach, which depends on in vivo synthesis of IL-10, requires much lower doses than systemic treatment. Neither mouse model mimics all aspects of human IBD, but such models are essential for development of new therapeutic approaches to IBD (13–20).

L. lactis is a nonpathogenic, noninvasive, noncolonizing Gram-positive bacterium, mainly used to produce fermented foods. We previously constructed recombinant L. lactisstrains for production and in vivo delivery of cytokines (21–23). We have now engineered an L. lactis strain (LL-mIL10) for secretion of biologically active murine IL-10 (mIL-10) (Fig. 1) (24, 25).

Figure 1

mIL-10 synthesis by LL-mIL10. (A) Western blot analysis of culture supernatant proteins from the mIL-10 producer strain LL-mIL10 (lane 1) and the vector control LL-TREX1 (lane 2) (24,25), revealed with anti–mIL-10 (Pepro Tech EC, London, UK). The position of mIL-10 is indicated. The concentration of mIL-10 in the culture supernatant was 3 μg/ml, as determined by enzyme-linked immunosorbent assay. The biological activity in the culture supernatant was estimated at 10,000 U/ml in a cell proliferation bioassay with the IL-10–dependent mast cell line MC/9 (33). When compared with a standard of known activity (BioSource International, Camarillo, California), the recombinant mIL-10 from the LL-mIL10culture supernatant revealed full specific biological activity. The NH2-terminus of this protein was determined, by automated Edman degradation, to be Gln-Tyr-Ser-Arg-Glu, which is identical to that of native mIL-10. (B) A 600(⧫), colony-forming units (CFU) (•), and mIL-10 concentration (▴) for an LL-mIL10 culture, prepared as for the inoculation of mice with 2 × 107 bacteria (24). Open symbols are for corresponding profiles of an identical culture that was UV-irradiated 3 hours after preparation. UV-irradiation immediately blocked the accumulation of mIL-10 in the culture supernatant and reduced the CFU count by 6 logarithmic units. No lysis was observed (25).

To evaluate the efficacy of the new therapeutic concept, we applied daily intragastric inocula (24) of 2 × 107 LL-mIL10 or control L. lactis(25) to mice in which chronic colitis had been induced by four cycles of administration of DSS in the drinking water for 7 days, alternating with 10-day periods of recovery (11,26). Treatment was arbitrarily initiated at day 21 after the fourth DSS administration. The predominant epithelial damage was loss of goblet cells and crypts (Fig. 2A) (25). A lymphocytic infiltrate was largely restricted to the thickened mucosa of the middle and distal colon; submucosa was affected in about 10% of mice. Inflammation typically persisted for at least 3 months. The chronic phase of inflammation showed complete regeneration of the intestinal epithelial lining. The range of inflammation is shown in histological images from the distal colon (Fig. 2A) and middle colon (25).

Figure 2

Intestinal histology of murine colitis models. Images represent sections of the distal colon (magnification, ×100). Hematoxylin and eosin staining. (A to C) Inflammation in the DSS-induced colitis model, established in female Balb/c mice. Lymphocytic infiltrate and disturbance of tissue architecture was observed in untreated mice (A). After treatment by 14 daily intragastric inocula of 2 × 107 LL-mIL10 per mouse followed by 14 days of recovery, the lymphocytic infiltrate was reduced and the tissue architecture was restored (B). Healthy control mice (C). (D to F) Images from 7-week-old female 129Sv/Ev IL-10−/− mice. Lymphocytic infiltrate is apparent in untreated mice (D) and mice treated for 4 weeks by daily intragastric inocula of 2 × 107 LL-TREX1 (E) but not in mice treated for 4 weeks by daily intragastric inocula of 2 × 107 LL-mIL10 (F).

Histological scores were devised to allow quantification of histological changes (Fig. 3A). Scores from individual mice (n = 10) after treatment with different L. lactis strains were recorded after blinded interpretation of sections from the distal colon (26). Untreated healthy mice had a histological score of 1, whereas mice with the induced chronic colitis and mock-treated control mice had a score of ∼5. Mice treated for 14 days with LL-mIL10 given by gastric catheter, followed by 14 days of recovery, had an average histological score of ∼3. This represents a nearly 50% decrease (P = 0.0151) in pathological symptoms, a more pronounced improvement than that obtained with standard systemic treatment with TNF monoclonal antibodies (mAbs), which leads to a 30 to 40% decrease in inflammation (26). In the LL-mIL10–treated group, colons of 4 of 10 mice had the histological characteristics of healthy mice, and 4 showed minor patchy remnants of inflammation (25); in 2 mice, larger areas remained affected. No further improvement was observed when LL-mIL10 treatment was extended to 4 weeks; when the daily inoculum was 109bacteria rather than 2 × 107, the improvement in pathology was less pronounced (27). Mice with DSS-induced colitis occasionally developed adenomas (25), corresponding to development in humans of adenocarcinoma often seen associated with ulcerative colitis. By contrast, no adenomas were seen in mice treated for 14 days with LL-mIL10.

Figure 3

Statistical evaluation of colon histology. Colon sections were randomly numbered and interpreted in a blinded manner. Scores from individual mice were subsequently decoded, and regrouped numbers were analyzed statistically. Bars represent the mean ± SEM. *P < 0.025; **P = 0.0151. (A) Histological scores (sum of epithelial damage and lymphoid infiltrate, both ranging from 0 to 4) for the distal colon of groups (n = 10) of control female Balb/c mice (white bar) and of female Balb/c mice with DSS-induced colitis that were untreated (hatched bar), treated with the indicated L. lactis cultures (black bars), or treated with five daily intraperitoneal injections of the compounds indicated (gray bars) (mIL-10: 5 μg per mouse per day; anti–IL-12: 1 mg per mouse per day; dexamethasone: 5 μg per mouse per day; rat IgG: 5 μg per mouse per day). Mice treated daily for 2 or 4 weeks (wk) with 2 × 107 mIL-10–producing LL-mIL10 showed significantly reduced inflammation when compared with untreated or control-treated (LL or LL-TREX1) mice. This effect was not observed when LL-mIL10 cultures were UV-killed (+UV). (B) Histological scores (sum of the degrees of inflammation in the proximal, middle, and distal colon, all ranging between 0 and 4) obtained after blinded interpretation of groups (n = 5) of 7-week-old untreated (hatched bar),LL-TREX1–treated, or LL-mIL10–treated female 129 Sv/Ev IL-10−/− mice (black bars).LL-mIL10–treated mice showed significantly less inflammation than untreated mice.

We also evaluated the ability of LL-mIL10 to prevent the onset of colitis in IL-10−/− mice, which spontaneously develop colitis at age 3 to 8 weeks (12). This progressive disease is characterized by multifocal inflammatory cell infiltrates (mononuclear cells and neutrophils), moderate epithelial hyperplasia, and slight mucin depletion from goblet cells. Other work has shown that treatment of 3-week-old IL-10−/− mice with recombinant mIL-10, interferon-γ mAbs, or IL-12 mAbs prevents the onset of colitis (16, 28). In our studies, the untreated IL-10−/− mice showed less severe inflammation than that observed by other workers (16, 29). We treated 3-week-old mice (n = 5) by daily intragastric inocula of 2 × 107 or 109 LL-mIL10or control L. lactis as for the DSS-induced colitis model (Fig. 2) (25). Untreated mice had a mean histological score of ∼4.5, whereas the LL-mIL10–treated group had a mean score of 1.5, which only slightly exceeds values reported for control mice (16) (Fig. 3B). In contrast to results with DSS-induced colitis, treatment efficacy was the same for both inocula concentrations (27).

We confirmed that the therapeutic effect was due to mIL-10 synthesized de novo by LL-mIL10 rather than to residual amounts of mIL-10 in the inocula. Indeed, because of the culture conditions used, a minor amount of mIL-10 (40 ng) (Fig. 1B) was present in the supernatant of the inoculum. The fate of this residual mIL-10 is likely acid denaturation, followed by breakdown in the stomach and duodenum (25). Diseased mice (DSS-induced colitis) treated for 2 or 4 weeks with ultraviolet (UV)-killedLL-mIL10 cultures (Fig. 1B) (25) showed no difference in colon histology compared with control mice positive for colitis (Fig. 3A). This result indicates that the therapeutic effects require physiologically active LL-mIL10.

We also investigated the synthesis of recombinant mIL-10 byLL-mIL10 in the intestine of IL-10−/− mice, which cannot themselves synthesize mIL-10. After administration of a total of 2.4 × 1010 LL-mIL10 (as serial inocula), we detected 7 × 108 LL-mIL10 and 7 ng of mIL-10 in the colon (25). Hence, these bacteria can actively produce mIL-10 in the colon, albeit at a lower yield than that observed in culture. This result agrees well with recent findings thatL. lactis is metabolically active in all compartments of the intestinal tract (30). Although LL-mIL10organisms were present in other areas of the gastrointestinal tract (cecum, ileum, jejunum, and stomach), mIL-10 was not detectable there. Perhaps mIL-10 reached detectable levels only in the colon because in this part of the intestine the protein is not degraded, and the contents move slowly enough to allow its accumulation.

We compared the performance of LL-mIL10–mediated mIL-10 delivery with that of standard anti-inflammatory methods: systemic treatment (five daily intraperitoneal injections) with recombinant mIL-10, IL-12 mAbs (31), or dexamethasone. All therapies decreased inflammation in DSS-induced colitis by ∼50% (Fig. 3A). Our method, however, required a much lower amount of mIL-10. We estimated that 14 daily inoculations of 2 × 107 LL-mIL10delivered ∼1 U of mIL-10 per mouse (25), i.e., an amount that is several orders of magnitudes lower than the optimized total amount of intraperitoneally injected mIL-10 (1.25 × 104 U per mouse).

We propose two possible routes by which mIL-10 might reach its therapeutic target. The lactococci may produce mIL-10 in the lumen, and the protein may diffuse to responsive cells in the epithelium or the lamina propria. Alternatively, the lactococci may be taken up by M cells (bacterial size and shape would allow this), and the major part of the effect may be due to recombinant mIL-10 production in situ in intestinal lymphoid tissue. Both routes may involve paracellular transport mechanisms that are enhanced in inflammation. After transport, mIL-10 may directly down-regulate inflammation. Alternatively, autocrine mIL-10 secretion by lymphoid cells, as shown by transfer of Tr1 cells (32), epithelial cells, or both, may be induced and may enhance repair.

In summary, the method described here—cost-effective localized delivery of a therapeutic agent that is actively synthesized in situ by food-grade bacteria—may have potential clinical applications for treatment of IBD, particularly as an alternative to systemic treatment. In principle, the method may also be useful for intestinal delivery of other protein therapeutics that are unstable or difficult to produce in large quantities.

  • * To whom correspondence should be addressed. E-mail: lothar.steidler{at}

  • These authors contributed equally to this work.


Stay Connected to Science

Navigate This Article