Discussion
Although intestinal dysbiosis is recognized as one of the pathogenic factors for the development of several pancreatic diseases, such as chronic alcoholic pancreatitis and pancreatic cancer [4, 5], pathogenic roles of intestinal dysbiosis in AIP have remained poorly defined. In our previous studies, we found that intestinal dysbiosis mediates experimental AIP through the activation of pDCs, which subsequently produce large amounts of IFN-α and IL-33 [20]. In this study, we demonstrated differences in the composition of fecal microbiota in samples from three type 1 AIP patients, at active and remitted phases. Interestingly, induction of clinical remission by PSL was accompanied by complete disappearance of Klebsiella spp. from the gut of two patients who harbored this bacterium before the treatment. We then explored the pathogenicity of Klebsiella spp.in experimental AIP by utilizing MRL/MpJ mice treated with 10 µg of poly (I:C). In contrast to the manifestations in the well-established severe AIP model induced by repeated IP injections of 100 µg of poly (I:C), repeated IP injections of poly (I:C) at a tenfold lower dose (10 µg) into MRL/MpJ mice resulted in the development of mild AIP [20]. By utilizing this mild AIP model, we found that the degree of AIP was greater in mice treated with both oral administration of heat-killedK. pneumoniae and IP injections of 10 µg of poly (I:C) than in those that received either treatment alone. Moreover, oral administration of heat-killed K. pneumoniae combined with IP injections of poly (I:C) promoted pancreatic accumulation of pDCs; consequently, this elevated IFN-α and IL-33 production by those cells. Thus, moderate AIP mediated by the pancreatic accumulation of pDCs was successfully induced in MRL/MpJ mice only when oral administration of heat-killed K. pneumoniae was combined with IP injections of poly (I:C). Therefore, oral administration of heat-killed K. pneumoniae increased the sensitivity to poly (I:C)-induced experimental AIP. Together with the complete disappearance of Klebsiella spp.from feces of type 1 AIP patients at the remission phase, these data strongly suggest that immune responses against Klebsiella spp.may be involved in the development of type 1 AIP.
Induction of remission by PSL led to complete disappearance ofKlebsiella spp. from the feces of two of the three type 1 AIP patients. In contrast, colonization by Ruminococcus spp. was promoted after PSL treatment in all of the three patients. Thus, this study suggests that gut colonization by Klebsiella spp. andRuminococcus spp. may play pro-inflammatory and anti-inflammatory roles, respectively, in type 1 AIP. Such increased abundance ofKlebsiella spp. and decreased abundance of Ruminococcus spp. have been observed in the feces of patients who developed pancreatic fistula after pancreaticoduodenectomy for pancreatic cancer [26]. In addition, fecal samples obtained from chronic alcoholic pancreatitis patients exhibited higher abundance of Klebsiella spp. than fecal samples from patients with alcoholic hepatitis [27]. These reports and the present study support the view thatKlebsiella spp. may function as a pathobiont for CP and AIP as in the case of primary sclerosing cholangitis (PSC) [28]. However, it should be noted that our sample size was too small to establish with certainty gut microbial alterations in type 1 AIP before and after PSL treatment. Moreover, pathogenic immune responses to colonization byKlebsiella spp. might be masked by the differences in microbiota profiles, as the proportions of Bacteroides, Streptococcus , andClostridium spp . were reported to be lower in patients with AIP than in those with CP [29].
Oral administration of heat-killed K. pneumoniae increased the sensitivity to experimental AIP induced by low doses of poly (I:C) in this study. Thus, immune responses activated by heat-killed K. pneumoniae alone were not sufficient for the induction of experimental AIP. As for the pathogenic intestinal bacteria other than K. pneumoniae , Haruta et al. reported that repeated IP injections of heat-killed Escherichia coli led to the development of experimental AIP in C57BL/6 mice [30]. They further identified FliC, one of the flagella subunit proteins of this bacterium, as a bacterial pathogenic factor because repeated IP injections of FliC alone caused AIP in C57BL/6 mice [31]. Therefore, persistent exposure to FliC derived from E. coli is sufficient for the development of experimental AIP in C57BL/6 mice. These previous studies utilizing heat-killed E. coli and C57BL/6 mice differed from our present data in that the development of AIP in MRL/MpJ mice requires administration not only of heat-killed K. pneumoniae but also of poly (I:C) at low doses. Whether persistent exposure to heat-killed intestinal bacteria plays a critical or supportive role in the development of AIP remains to be elucidated. Nonetheless, those studies and our present data indicate that the development of immune reactions against intestinal bacteria may affect AIP.
One question arising from the present study is how heat-killed K. pneumoniae activates pDCs, which produce large amounts of IFN-α and IL-33. In this regard, host defense against pulmonary infection withK. pneumoniae depends upon TLR9-mediated proinflammatory cytokine responses [32]. Moreover, the depletion of pDCs impairs proinflammatory cytokine responses in mice during sublethal pulmonary infection with K. pneumoniae [33]. Given the fact that TLR9 activation is a strong inducer of IFN-α production in pDCs [34, 35], it is possible that heat-killed K. pneumoniae directly stimulates IFN-α production by pDCs through TLR9 activation. This idea is supported by the fact that activation of type I IFN signaling pathways, and subsequent production of C-X-C motif chemokine ligand 10, mediate host defense against murine K. pneumoniae pneumonia [36, 37].
Nakamoto et al. identified K. pneumoniae as one of gut pathobionts for PSC [28]. Colonization by K. pneumoniaederived from PSC patients into gnotobiotic mice led to the development of hepatobiliary injury through the induction of CD4+T cells, which produced IFN-γ and IL-17 [28]. Colonization byK. pneumoniae increased the sensitivity to experimental PSC, as the development of chronic hepatobiliary inflammation required both inoculation of this bacterium and oral administration of 3,5-dicarbethoxy-1,4-dihydrocollidine, the latter of which triggered hepatobiliary injury. Together with our findings, those data suggest that immune responses against K. pneumoniae may promote the development of AIP and PSC in the presence of potent triggers. It should be noted, however, that AIP and PSC are completely different in terms of the sensitivity to PSL treatment in that most of patients with type 1 AIP respond to PSL treatment while those of PSC do not. Therefore, other factors than the pathogenic colonization of Klebsiella spp. might be involved in the different pathophysiology between these immune disorders.
In conclusion, colonization by Klebsiella spp. may mediate the development of type 1 AIP through the activation of pDCs.Klebsiella spp. may be a type 1 AIP pathobiont. However, a confirmation of this hypothesis awaits future studies in which gut microbiome will have to be studied in a large number of type 1 AIP patients.