LIPID A-MEDIATED TOLERANCE AND CANCER THERAPY

Early, late, and cross tolerance

The occurrence of tolerance or host unresponsiveness in animals and humans administered multiple doses of microbe or microbial products has long been recognized by scientists and physicians with published reports appearing in professional journals dating back to the 19^th century.¹ Many of the very early observations focused largely upon the establishment of pyrogen tolerance in which animals treated with a microbe or a microbial product exhibited a refractory state for the development of fever, or a markedly diminished fever response, upon subsequent treatments with the same or a related microbe or product. Following the identification and purification of lipopolysaccharide (LPS) in the 20^th century, it was determined that the microbial product, LPS itself, can be highly tolerogenic.² In this respect, many of the pleiotropic effects of LPS, including fever, induction of cytokine production, and even mortality are absent or markedly diminished upon repeated administration of LPS. From these observations, it has been postulated that tolerance serves to protect the host from the detrimental consequences of the robust and extensive inflammatory responses that follow exposure to LPS. Tolerance cannot, however, be characterized as a global downregulation of responsiveness as some LPS-responsive characteristics remain unchanged or in some cases can actually be upregulated in experimental models of tolerance.

In studies carried out by numerous investigators over the years to clarify the concept of tolerance, a variety of terms have been used to designate the observed alteration in immune responses that occur upon repeated administration of LPS. Many alternative terminologies in lieu of the word “tolerance” have been suggested, these include reprogramming, deactivation, adaptation, refractoriness, hyporesponsiveness, and desensitization. All are arguably accurate to some extent, although many of these should not be used interchangeably, as some of these terms refer to tolerance that can be established only under specific conditions.

Numerous studies have been dedicated to the elucidation of LPS structures from various microbes in order to account at the molecular level for LPS tolerance. From these studies, it has been established that the lipid A moiety of LPS is responsible for almost all of the described activities of LPS.^3-8 Accordingly, the induction of tolerance is also routinely observed with repeated administration of highly purified lipid A; responses to which are virtually indistinguishable from those observed with repeated LPS administration.⁹ In addition to tolerance to repeated lipid A/LPS treatment, cross-tolerance can be obtained in which other microbial products, such as lipoteichoic acid, induce tolerance to the effects of lipid A/LPS and vice versa.^10-12 The concept of tolerance is also distinguished temporally on the basis of the time required to establish this phenomenon. In this regard, both early and late tolerance have been described. Early tolerance is usually observed within hours of LPS administration and may last up to a week or more. During early tolerance, suppression of cytokine production, fever, and endotoxin lethality occur independently of antibody production.^13-15 Cross-tolerance is generally associated with early, but not late tolerance. In contrast to early tolerance, late tolerance is induced many days after LPS treatment when early tolerance has generally ceased or markedly waned. Late tolerance is characterized by the production of antibodies directed against LPS, which confers the specificity and lack of cross-tolerance observed in this phase. The following discussion will be driven primarily toward an examination of the consequences of early lipid A/LPS tolerance.

The consequences of the development of early lipid A/LPS tolerance on the therapeutic use of lipid A is two-fold. While subtoxic doses of lipid A may be used to either mitigate the side effects of lipid A/LPS treatment and/or protect against endotoxin shock and thus may be considered beneficial, the diminished therapeutic efficacy of lipid A in repeated administration over time may also be considered a deleterious effect. Consequently, a number of lipid A analogs have been developed to both decrease tolerance in order to increase efficacy as well as to increase tolerance to either diminish the associated toxicity of lipid A/LPS treatment or protect against endotoxin shock in patients susceptible to sepsis.^9,16-20

The isolation of various lipid A structures and synthesis of analogs

An extensive array of lipid A molecules and structurally-related analogs have been isolated and/or synthesized in an effort to identify molecules that could act as competitive inhibitors and/or be used to reduce the toxicity of LPS. Indeed, comparisons of the biological potencies of lipid A structures derived from different bacterial strains and methods of preparation have been reported over the past several decades.^21-27 Monophosphoryl lipid A (MPL) was among the first structures to be purified and structurally characterized by our laboratory.^24,27 The results of these seminal studies led to the initial determination of the complete structure of lipid A. The subsequent development of a number of highly purified MPL analogs (with respect to number of fatty acyl groups) led to the analysis of numerous analogs in regard to toxicity and efficacy in tumor regression. MPL was found to be generally less toxic than diphosphoryl lipid A and intact LPS, which generated interest in its potential therapeutic use.^26,27 The relative efficacy and toxicity of MPL is, however, somewhat dependent upon the purity of the preparation and the strain of bacteria from which it is derived. In general, hexaacylated lipid A structures were shown to be more toxic than the pentacylated or tetraacylated structures and the chain length of the fatty acyl groups was also considered an important factor in toxicity.¹⁸ Although generally less toxic than most diphosphoryl lipid A structures and LPS, MPL exhibits some agonist activity, albeit at a markedly lower level than that of intact LPS.²⁸ In contrast, the pentaacyl diphosphoryl lipid A derived from LPS of Rhodobacter sphaeroides (RsDPLA) was purified and characterized in our laboratory for its ability to act as a nontoxic effective antagonist of LPS and agonist lipid A moieties in both human and murine cells.^29-33 In addition to native lipid A structures, a relatively large number of synthetic lipid A analogs were later developed and screened for toxicity as well as LPS-mimetic or antagonistic activity.^20,34,35

Relevance of tolerance to the use of LPS/lipid A in cancer

The phenomenon of tolerance in immunotherapy for the treatment of cancer has been observed as long as LPS/lipid A has been used as a cancer therapeutic. Indeed, to avoid the induction of tolerance, the late nineteenth century physician, Dr. William Coley, found it necessary to use incremental doses of a toxin formulation prepared from culture filtrates of bacteria in treating his sarcoma and carcinoma patients. “Coley’s toxin”, as the aforementioned preparation came to be called, was found to have mixed success, but was used for many years by physicians for a variety of different malignancies. The tolerance induced by Coley’s toxin may have been the result of LPS/lipid A tolerance, cross tolerance, or a combination of both as Coley’s toxin was comprised of killed bacteria of both gram-positive and gram-negative strains.¹ While Coley’s toxin was comprised of a variety of microbial products that included LPS, LPS alone was also found to cause tumor regression.² It was later determined that LPS-induced hemorrhagic necrosis of tumors is primarily due to the induction of a serum factor, termed tumor necrosis factor (TNF).³⁶ Further investigation revealed that the lipid A portion of LPS was primarily responsible for the induction of TNFα.³⁷ In the late twentieth century, the results of several clinical trials using LPS as a therapy were reported in cancer patients (Table 1). While purified LPS was confirmed to have positive antitumor activity in humans, both the toxicity of LPS as well as the relatively rapid induction of tolerance by LPS detracted from its overall utility as a cancer chemotherapeutic. The decreased antitumor activity of LPS due to tolerance was similar to the reduced antitumor activity observed with multiple administrations of TNFα, suggesting that the tolerance observed in vivo may well be due to both reduced TNFα activity as well as the diminished induction of TNFα by repeated LPS administration.³⁸

Table 1.

The role of tolerance in the use of various lipid A moieties in cancer.

Lipid A Moiety	Activity	Role of tolerance
LPS (lipid A)	Tumor regression/antimetastasis	Detrimental: diminishes efficacy
Tolerin	Protection against sepsis	Beneficial: prevents endotoxin shock
MPL/Detox	Adjuvant for cancer vaccine	Unknown
ONO-4007	Tumor regression	Detrimental, however induces less tolerance than LPS
DT-5461	Tumor regression/antimetastasis, protection against sepsis	Detrimental for anti-tumor activity, but beneficial for prevention of LPS shock
SDZ MRL 953	Protection against sepsis	Beneficial: prevents endotoxin shock

LPS in clinical and animal studies

As discussed previously, one of the primary problems with the use of LPS/lipid A as a therapeutic intervention in the treatment of cancer is the rapid induction of tolerance that diminishes the efficacy of the treatment. To address the problem of decreased antitumor activity of LPS after multiple administrations, Mackensen et al. evaluated the effect of endotoxin tolerance upon cytokine production in cancer patients following repeated daily LPS injections. Patients treated a second time with LPS, 24 hours after an initial LPS treatment, had significantly decreased levels of circulating TNFα, interleukin (IL)-6, IL-8, granulocyte colony stimulating factor (G-CSF), and macrophage colony stimulating factor (M-CSF). The levels of most of these mediators continued to decline to baseline levels upon subsequent daily injections of LPS for 5 days. Interestingly, IL-6 levels plateaued upon the third day of LPS injections with no further decline in days 4 and 5. The authors also reported that side effects, such as fever and chills, increased on the second day of LPS treatment, which did not correlate with the reported decreases in cytokine levels. In addition to tolerance induced by daily LPS injections, the authors also noted that tolerance can be induced with repeated LPS injections of 1 week and 2 week intervals as well.³⁹

In initial clinical trials carried out by the same group to evaluate LPS therapeutic benefit, induction of tolerance was reduced by using escalating doses of LPS.⁴⁰ In a subsequent phase I trial, tolerance was further reduced through the use of both escalating doses of LPS as well as an increased treatment interval of two weeks, which results in less tolerance induction than shorter treatment intervals. While moderate or strong antitumor activity with this treatment regimen was observed in a few of the reported cases, most cases did not demonstrate significant antitumor effects .⁴¹ The authors hypothesized that the relatively limited antitumor effect of LPS was, at least in part, due to the induction of LPS tolerance.^42,43

The same group also conducted a phase I clinical study in which interferon γ (IFNγ) was administered in addition to LPS, in an effort to prevent LPS tolerance and thereby increase the therapeutic efficacy of the treatment. Pretreatment with IFNγ not only prevented LPS tolerance induction but, in fact, induced higher levels of TNFα, IL-6, and G-CSF than in the initial administration with LPS alone. Conversely, the downregulation of IL-8 upon repeated LPS administration was unaffected by IFNγ pretreatment. Whether the diminished LPS tolerance observed with IFNγ pretreatment correlated with improved antitumor activity was not reported, however.⁴⁴

In addition to clinical trials in humans, the induction of LPS/lipid A-mediated tolerance has also been widely examined in experimental animal models. Numerous tactics have been found to delay or prevent tolerance to LPS/lipid A in animals, including treatment with recombinant IFNβ, nitric oxide synthase inhibitors, p38/stress-activated protein kinase-2 inhibitors, administration of flt3 ligand (a growth factor important for dendritic cell differentiation), and many other methods as well.^45-48 Interestingly, one group recently attempted to counteract the effects of LPS/lipid A-mediated tolerance by using LPS in combination with cytotoxic drugs, such as 4’-(9-acridinylamino)-methansulfone-m-aniside, cyclophosphamide, 1-octadecyl-2-methoxy-rac-glycero-3-phosphocholine, and hexadecylphosphocholine. The authors reported, however, that they were unable to produce sufficient anticancer therapy with acceptable toxicity using this approach.⁴⁹

While the anticancer activity of LPS in its native form has been extensively investigated, the potential therapeutic efficacy of irradiated LPS in cancer patients has also been reported. Presumably, the irradiation of LPS produces a variety of different lipid A structures. The overall goal of clinical studies with irradiated LPS, which is currently registered under the market name of Tolerin®, was somewhat different than the aforementioned studies with native LPS. Tolerin was designed to be administered with the purpose of inducing tolerance to LPS and at the same time, boosting natural immunity with the intent of protecting highly susceptible immunosuppressed cancer patients from sepsis and subsequent lethal septic shock. The results from these recently published studies have documented that Tolerin was well-tolerated by the patients and increased natural resistance which, importantly, correlated with a decreased incidence of infection in these patients.⁵⁰

Although the majority of studies that have investigated LPS/lipid A in the treatment of cancer have focused upon their effects on inhibition of tumor growth and/or protection against sepsis, other studies have been directed toward the reverse, namely the effects of tumors upon endotoxin lethality. Interestingly, Berendt et al. reported an increased sensitivity to endotoxin in mice bearing LPS-sensitive tumors, which the authors maintain is analogous to increased endotoxin lethality in mice infected with pathogens that cause systemic macrophage activation. Furthermore, the authors report systemic activation of macrophages in mice bearing LPS-sensitive tumors which they correlate to the increased endotoxin lethality observed in these animals.⁵¹ In contrast, numerous studies have reported that many tumors are capable of inducing a tolerant state in immune cells.^52-55 While the term “tolerance” is routinely used to describe both the hyporesponsive state of leukocytes induced by tumor cells as well as the diminished cytokine production, fever, and lethality following multiple administrations of LPS/lipid A, these are two separate phenomena. Although there may be similarities between these two different types of tolerance, the two phenomena can be distinguished from one another by the causative agents, which are presumably one or more factors produced by tumor cells and microbial products, respectively. In addition to these direct effects of the tumors themselves upon the ability of immune cells to respond to LPS, surgical removal of tumors can also impact responsiveness of immune cells to LPS. A general state of immunosuppression is routinely observed in humans and animals following surgery.⁵⁶ Moreover, it has been reported recently that cryosurgery of tumor tissue induces tolerance to endotoxin-mediated lethality, suggesting that this surgical procedure may also be protective against septic shock in cancer patients.⁵⁷ Collectively, these studies suggest that the effects of tumors upon immune cell activation overall, and LPS responsiveness specifically, appear to be complex and seem to differ based upon the nature of the tumor. In addition, surgical removal of tumors may well contribute to the induction of tolerance to LPS/lipid A, which may have the dual effect of diminishing the efficacy of lipid A treatments as well as protecting the patient from septic shock.

Monophosphoryl lipid A and lipid A adjuvants

Following the isolation, derivatization, and synthesis of various lipid A structures, it was quickly established that some of the beneficial effects of the identified nontoxic lipid A moieties had therapeutic potential. For instance, SDZ MRL 953 and MPL were found to enhance host defenses against subsequent bacterial infection through induction of G-CSF, M-CSF, and other cytokines.^26,58-60 At the same time, however, these compounds induced refractoriness (tolerance) to LPS toxicity through reduced secretion of inflammatory cytokines, while host cell phagocytic activity was maintained. Some of the nontoxic lipid A moieties, such as MPL, also demonstrated tumor regression activity.⁶¹ Similar to LPS, however, MPL has also been shown to induce tolerance in both experimental animals as well as in patients enrolled in clinical trials.^26,28,62-64 Due to these confounding factors, MPL is not currently in use clinically for its tumor regression activity per se, but remains important in cancer therapy as an immunoadjuvant in cancer vaccines (see chap. 10).⁶⁵ Because macrophage activation is critical to the adjuvant activity of lipid A (and in the development of tolerance), lipid A adjuvancy may also be susceptible to tolerance, depending upon experimental conditions and treatment administration.

Synthetic lipid A analog, ONO-4007

In addition to the various native forms of lipid A derived from different bacterial strains, numerous lipid A analogs have been synthesized and screened for antitumor activity. Several synthetic lipid A analogs have demonstrated potential as cancer chemotherapeutics, but the induction of tolerance upon repeated administration has only been reported for a fraction of these. ONO-4007, a synthetic triacylated monosaccharidic lipid A analog, has been the focus of numerous studies in both animals and humans, which have shown that it exhibits lower toxicity than LPS and causes tumor regression.^66-68 While ONO-4007 has been shown to induce tolerance in animals using several different models, it induces less tolerance than that induced by LPS and synthetic E. coli-type lipid A (LA-15-PP).^17,69,70 Interestingly, differential tolerance to ONO-4007 was observed in different tissue types. While significant tolerance to ONO-4007-mediated cytokine induction was observed in serum and liver tissue, no tolerance was observed in tumor tissue extracts as assessed by TNFα production in tumor-bearing mice treated with ONO-4007 at 8, 12, and 15 days following tumor implantation.¹⁷

The results of further investigations have established that the tumor tissue did become hyporesponsive to stimulation with ONO-4007 as measured by TNFα induction, but tumoral responses recovered more quickly than responses in liver and serum and were completely responsive by 72 hours. Similarly, tumor infiltrating macrophages recovered from ONO-4007-mediated hyporesponsiveness within 72 hours after initial exposure. The authors hypothesized that the selective recovery of tumor tissues may, at least in part, be due to constant recruitment of macrophages to tumor tissue. In addition, the authors demonstrate that repeated injections of LA-15-PP into mice enhanced its clearance from blood circulation, whereas the clearance of ONO-4007 was stable even following multiple administrations, suggesting that pharmacokinetics also play a role in the differences in tolerance between these two lipid A structures. Moreover, TNFα tissue levels peaked 1-2 hours following ONO-4007 treatment and then decreased in the spleen and liver, but remained elevated for at least 6 hours in tumor tissue.^71,72

Synthetic compound, DT-5461

Like LPS and ONO-4007, tolerance to the synthetic tetraacylated lipid A analog, DT-5461, is also observed. Tolerance to DT-5451 occurs one day after treatment with a return to responsiveness observed 3-5 days later.⁷³ In experimental animal models of cancer using prostaglandin E₂ (PGE₂)-producing tumors, a combined treatment of indomethacin and DT-5461 was shown to have significant antitumor activity and an additive effect upon survival. The authors have hypothesized that the antitumor effect of the combined therapy is due to a combination of TNFα activity, as well as the inhibition of PGE₂ production.⁷⁴ Interestingly, PGE₂ has also been hypothesized to be a mediator of LPS/lipid A-induced tolerance.⁷⁵ Furthermore, and of some interest, cyclooxygenase inhibitors have been reported to prevent the induction of tolerance to LPS.⁷⁶

While the majority of studies of DT-5461 have focused upon its antitumor activity, DT-5461 has also been shown to be protective against endotoxemia. DT-5461 pretreatment induced significant tolerance to lethal LPS exposure in mice.⁷⁷ Similar to studies described earlier with Tolerin, the induction of tolerance by DT-5461 was found to be beneficial therapeutically for the protection of immunosuppressed patients from sepsis, including cancer patients undergoing chemotherapy or radiation treatment. In addition to the induction of tolerance by DT-5461 itself, LPS also induces tolerance to DT-5461 as well, strongly suggesting that these two compounds act through similar cellular mechanisms. In mice injected with LPS at daily intervals for a week, the antimetastatic activity of DT-5461 was significantly reduced, further supporting the conclusion that the mechanisms of tolerance by LPS and DT-5461 are identical or cross-tolerance occurs between the two mediators.⁷⁸

Synthetic compound, SDZ MRL 953

Similar to Tolerin and DT-5461, the LPS tolerance induced by the synthetic triacylated lipid A analog, SDZ MRL 953, has been evaluated for its protective effects against sepsis in cancer patients.⁷⁹ SDZ MRL 953 pretreatment inhibited LPS-induced TNFα, IL-1β, IL-8, IL-6, and G-CSF serum levels, suggesting that it induces tolerance to LPS and might be effective as a prophylactic treatment for patients who may otherwise be susceptible to sepsis.⁷⁹ Further clinical studies will be needed, however, to determine the overall efficacy of SDZ MRL 953 in the prevention of gram-negative infections and septic shock in immunosuppressed cancer patients.

Mechanisms of early LPS/lipid A-mediated tolerance

Because many investigators have demonstrated that lipid A is the active component of LPS, and is also capable of inducing LPS tolerance, presumably most of the LPS-mediated activity that occurs during early tolerance can be attributed to the lipid A portion of the molecule.^28,80,81 These conclusions are also strongly supported by the results of studies using a variety of lipid A analogs as described in previous sections. Despite hundreds of papers that have been published concerning LPS/lipid A-mediated early tolerance, the underlying mechanism still remains to be determined. Indeed, numerous hypotheses and supportive studies suggest that multiple mechanisms may play a role in tolerance. If multiple pathways do exist, they may be activated concurrently or at temporally distinct points in the activation/deactivation pathway sequences. Alternatively, different mechanisms may operate in a manner that is distinct from one another. The activation of one particular mechanism rather than another may depend upon the specific cell type involved, the local environment within the host, or the experimental conditions in vitro. While the majority of studies of early tolerance to LPS/lipid A have focused upon macrophages, a number of other cell types have also been shown to be susceptible to tolerance.^82,83 Because detailed reviews of the numerous studies focusing upon the mechanisms of tolerance have recently been published, only a broad overview will be included in this chapter. The reader is referred to a number of excellent reviews for a more comprehensive discussion concerning the role of different LPS/lipid A signaling pathways in the induction of tolerance.^13,75,84,85

Toll-like receptors, associated signaling molecules, and negative regulators

The members of the toll-like receptor (TLR) family are unique in their ability to recognize pathogen associated molecular patterns (PAMPs) found in variety of microbial products, including LPS/lipid A. LPS has been shown to activate cells through the TLR4 pathway and to a lesser extent, through TLR2. CD14 and MD-2 are both proteins that have been shown to associate with, and also be required for, TLR4-dependent activity and are therefore important for activation of cells by LPS/lipid A (Fig.1). Upon activation of TLR4 by LPS, the adapter protein, myeloid differentiation factor 88 (MyD88) is recruited to the cytoplasmic domain of TLR4, where it associates with and then activates the IL-1 receptor associated kinase-1 (IRAK-1). Activated IRAK-1 then dissociates from the TLR4 complex and subsequently binds and activates TNF receptor associated factor-6 (TRAF-6), which in turn activates TGFβ-activated kinase-1 (TAK1) and nuclear factor of κB-inducing kinase (NIK). The activation of TAK1 and NIK results in the activation of the mitogen activated protein kinase (MAPK) and nuclear factor of κB (NFκB) pathways. ^75,85 The reader is referred to previous chapters of this volume for more information concerning the signaling pathways of lipid A/LPS.

Fig. 1.

Schematic diagram of hypothetical mechanisms of tolerance within the TLR4 pathway. While ST2, IRAK-M, and the splice variant of MyD88 (MyD88s) are thought to suppress association of IRAK-1 with MyD88 through competitive inhibition. TOLLIP has been shown to bind to IRAK-1 and inhibits its activity. SOCS-1 and SOCS-3 are inhibitors of the JAK/STAT pathway. While the mechanism by which SHIP induces tolerance is unknown, it is believed to inhibit the NFκB signaling pathway.

Numerous studies have reported the role of TLRs and associated proteins in the induction of tolerance. Although it would be intuitive that downregulation of TLR4 or another component of the TLR4 receptor complex may be responsible for LPS/lipid A-induced tolerance, this remains a somewhat controversial issue. While some studies have reported a downregulation of TLR4 receptor protein expression following LPS treatment, other studies have demonstrated that TLR4 expression is either unaffected or sometimes even increased upon LPS administration.^83,86-93 In contrast to TLR4, studies of the effects of LPS on TLR2 expression have been more conclusive, with no downregulation of TLR2 observed in most models of tolerance.^87-91 In addition, the expression levels of CD14 and MD-2 have also been evaluated in numerous studies. The results of most studies examining CD14 expression have shown no decrease in protein or transcript levels following LPS treatment.^88,91,94-96 Similar to TLR4, however, the results with MD-2 have been inconsistent. While many studies demonstrate little change in MD-2 expression with LPS/lipid A treatment, some studies show decreased MD-2 transcription and surface expression of the TLR4/MD-2 complex.^89,91,94,97 Interestingly, tolerance has been demonstrated in HEK293T cells in which TLR4, CD14 and MD-2 have all been overexpressed, suggesting that decreased CD14/TLR4/MD-2 expression is not required for the induction of tolerance in this model.⁹⁸ Overall, these studies suggest that, while there may be decreased expression of TLR4/MD-2 upon LPS/lipid A administration under certain experimental conditions, it is by no means a universal requirement for tolerance.

The role of a number of different signaling molecules downstream of the CD14/TLR4/MD-2 complex has also been investigated. Although expression levels of MyD88 have not been found to be decreased in tolerant monocytes, a marked inhibition of MyD88/TLR4 association has been reported.⁹⁰ In addition, a splice variant of MyD88 that is expressed upon LPS treatment, and functions to inhibit LPS signaling, has garnered considerable attention as a potential mechanism of tolerance induction.^99,100 Similarly, IRAK-1 activity and MyD88/IRAK-1 association have been reported to be diminished in LPS-tolerant cells.^90,101,102 Moreover, upregulation of the inactive kinase, IRAK-M, may also play a role in the induction of tolerance in monocytes. IRAK-M is markedly upregulated upon LPS treatment, causing inhibition of LPS signaling, possibly through competitive inhibition of IRAK-1 binding to MyD88. Of particular relevance, the induction of tolerance is significantly diminished in macrophages lacking IRAK-M as compared to macrophages derived from wild-type mice.^103,104

In addition to IRAK-M, other negative regulators of LPS signaling have also been investigated. Suppressor of cytokine signaling-1 (SOCS-1) and SOCS-3 have been shown to be rapidly induced upon LPS exposure, and these proteins can serve as negative feedback regulators of the janus activated kinase (JAK)/signal transducers and activators of transcription (STAT) signaling cascade, which is the downstream mechanism of signal transduction of many cytokine receptors.^105,106 The results of studies examining the role of SOCS-1 in tolerance have once again been mixed, with one report demonstrating an absence of tolerance to endotoxin lethality and LPS-mediated TNFα secretion in SOCS-1 -/- mice, while a different group demonstrated no difference in the induction of LPS tolerance to IL-12 secretion in SOCS-1 -/- macrophages. As a consequence, the role of SOCS-1 in tolerance remains inconclusive.^107,108

ST2, a member of the Toll-IL1 superfamily, is another inhibitor of LPS signaling. There is evidence to suggest that ST2 suppresses signaling of both the TLR4 and the IL-1 receptor through the sequestration of MyD88 and Mal adaptor proteins. Although the role of ST2 in tolerance has not yet been extensively investigated, recent reports demonstrating a failure of ST2-deficient mice to develop tolerance to endotoxin lethality, as well as LPS-mediated IL-6 and IL-12 production, suggest that ST2 may also play a role in the induction of tolerance.^109,110 These promising preliminary findings are likely to provide the basis for additional study of mechanisms of LPS/lipid A-mediated tolerance induction.

SH₂-containing inositol phosphatase (SHIP) and toll interacting protein (TOLLIP) are also negative regulators of LPS signaling. While TOLLIP mRNA and protein levels have been reported to be upregulated in LPS-tolerant cells, the role of this protein in tolerance has not been extensively investigated.^111,112 Interestingly, the results of recent studies with TOLLIP null mice indicate that TOLLIP can also act as an activator of LPS signaling, at least under certain circumstances.¹¹³ In comparison to TOLLIP, there is stronger evidence for involvement of SHIP in endotoxin tolerance. SHIP protein levels are significantly elevated following LPS treatment, a response which is mediated by LPS-induced TGFβ secretion. Accordingly, treatment with antibodies against TGFβ blocks LPS-induced tolerance. Notably, LPS-mediated tolerance cannot be induced in SHIP null mice.¹¹⁴

The suppressive effects of NO on cytokine production, as well as on immune cell proliferation and growth, have been extensively described.¹¹⁵ As a result, the effect of NO upon tolerance has also been investigated. Supportive evidence for NO involvement in the induction of tolerance includes increased NO production by tolerant peritoneal macrophages, inhibition of tolerance to endotoxin lethality by NO synthase inhibitors, and increased survival to a lethal dose of LPS in rats treated with an NO donor.^116-120 While many studies report increased NO production and/or iNOS transcription in tolerant macrophages, other studies have shown decreased NO production and NO synthase activity by macrophages in experimental models of tolerance.¹²¹ Similarly, and perhaps not totally unexpectedly, contradictory evidence for the role of NO in tolerance is also found in investigations with knockout mice. While Dias et al. have recently reported that iNOS null mice do not become tolerant to LPS-mediated pyrogenicity, Zingarelli et al. earlier demonstrated tolerance to LPS-mediated lethality and TNFα production in iNOS null mice.^122,123 The differences between the two studies may be dependent upon the different doses of LPS used, the endpoints measured, or the treatment regimens which also differed substantially between the two sets of experiments.

Corticosteroids, anti-inflammatory cytokines, and prostaglandins

In addition to participation of a number of components of the TLR4 pathway and the negative regulators associated with this pathway in the development of tolerance to LPS/lipid A, other mechanisms have also been proposed to mediate this phenomenon. LPS-induced glucocorticoid secretion has been well-documented and results in the suppression of a variety of different immune cell types. As such, the role of glucocorticoids as potential mediators of tolerance has remained an area of interest for decades. Despite numerous studies to address this hypothesis, however, strong evidence for the involvement of glucocorticoids in the induction of tolerance has been relatively sparce.¹³ While currently inconclusive, the role of glucocorticoids in tolerance continues to be investigated.

Anti-inflammatory cytokines, such as IL-10 and TGFβ, have also been suggested to contribute to the induction of tolerance to LPS/lipid A treatment. Support for this concept comes from the findings of studies showing significant downregulation of TNFα and other proinflammatory mediators by IL-10.^124,125 Furthermore, treatment with an antibody specific for IL-10 results in an inhibition of the induction of LPS/lipid A-mediated tolerance.¹²⁶ Conversely, tolerance to LPS/lipid A treatment can be readily induced in IL-10 knockout mice, suggesting that IL-10 cannot be the sole mediator of tolerance.¹²⁷ Indeed, induction of IL-10 by LPS itself is diminished upon subsequent administration of LPS under certain experimental conditions.¹²⁶ Similar to IL-10, pretreatment of isolated human peripheral blood mononuclear cells with recombinant TGFβ also suppresses LPS-mediated TNFα secretion, and a blocking antibody against TGFβ also diminishes the development of tolerance to LPS.^126,128 In addition to immunosuppressive cytokines, the synthesis and release of immunosuppressive prostaglandins, such as PGE₂, have also been hypothesized to contribute to the induction of tolerance in some experimental models. The basis for this hypothesis is that PGE₂ levels have been shown to be highly elevated in tolerant cells reexposed to endotoxin in a number of different animal and human models. Moreover, PGE₂ inhibits cytokine production in activated macrophages and lymphocytes.^129-132 Further studies are needed to more completely elucidate the role of PGE₂ in the induction of tolerance, however.

Transcriptional mediators

Peroxisome proliferator activated receptor γ (PPARγ) activation has been correlated with inhibition of macrophage activation by LPS, as assessed by cytokine and NO production.^133-135 Consequently, the role of PPARγ in LPS-mediated tolerance has been examined. Increased PPARγ binding to the PPAR response element (PPRE) and PPARγ transcriptional activity have been observed in LPS-tolerant cells, suggesting possible involvement of PPARγ in endotoxin tolerance.¹³⁶ Furthermore, the use of blocking oligonucleotides for the PPRE inhibited tolerance induced by LPS. Collectively, these preliminary studies suggest a potential role for PPARγ in tolerance, but more extensive investigations are required before definitive conclusions can be reached.

In part because so many LPS-responsive genes have been shown to be regulated by NFκB, there has been considerable interest in the numerous reports of decreased NFκB activation in the development of LPS tolerance, which has been widely observed in a variety of experimental models of tolerance. The results of some of these studies suggest that the decrease in NFκB activity may be due to an increase in the formation of p50/p50 homodimers, which are transactivationally inactive and therefore would be expected to antagonize the active p50/p65 heterodimer by competitively binding to κB sites in the promoters of LPS-responsive genes.⁹⁵ This hypothesis is supported by the reportedly increased levels of p50/p50 homodimers in tolerant cells and the failure to induce tolerance in p50-deficient mice.¹³⁷ Contradictory evidence has, however, also been presented, demonstrating that LPS tolerance could still be induced in p50 -/- mice, as assessed by IL-12 and TNFα production by splenocytes.¹³⁸ Collectively, these data suggest that, while decreased NFκB transcriptional activity is likely to be a causative factor in the induction of tolerance in a number of different experimental models, current data suggest that p50/p50 homodimers are not likely to be the only mechanism responsible for the diminished NFκB activity in various models of LPS tolerance.

The role of the proteasome

Our recent data have provided convincing evidence for a prominent role of the proteasome, a cytoplasmic organelle with multiple protease activities, in LPS signaling and subsequent development of inflammatory and immune responses. Structurally, proteasomes exist as multi-subunit complexes, consisting of a number of distinct, well-characterized, proteins.¹³⁹ The so-called 26S proteasome complex (2.5 MDa) is comprised of a 20S proteasome, which exhibits proteolytic activity, and a 19S proteasome, which provides regulatory functions.¹⁴⁰ The 20S proteasome has been defined structurally as a hollow, cylindrical, multi-protein structure composed of 28 protein subunits that are derived from 14 distinct gene products.^141,142 The protein subunits of the proteasome are arranged in four heptameric rings shaped approximately as a barrel. The three proteases of the proteasome are X (LMP7) (chymotrypsin-like protease activity), Y (LMP2) (post-glutamase protease activity), and Z (MECL-1) (trypsin-like protease activity) and these have been described in detail.^142,143 The protease activities of the proteasome have been shown to be regulated by IFNγ. Subunits LMP7, LMP2, and MECL-1 of the 20S proteasome are recognized as IFNγ-inducible proteasome-associated β-subunits. There is an overproduction of these subunits due to IFNγ produced early on during an inflammatory response, resulting in the introduction of the subunits into newly assembled proteasomes, which have been termed immunoproteasomes. Immunoproteasomes appear to have enhanced capability for generating class I MHC-binding peptides, as compared with “standard” proteasomes, cleaving more efficiently after basic or hydrophobic residues and less efficiently after acidic residues.¹⁴⁴

The role of the proteasome in LPS-induced inflammation had not been extensively pursued until our demonstration that LPS binds specifically to A1 (C2) and B4 (N3) proteins of the 20S proteasome complex.¹⁴⁵ After demonstrating that LPS binds proteasome subunits, we then assessed the potential physiological relevance of these interactions. To this end, we first carried out studies to determine the extent to which LPS modulates the proteasome’s proteolytic activity. We demonstrated that the addition of LPS to partially purified proteasomes in vitro activated the chymotrypsin-like and post-glutamase activities of macrophage proteasomes.^145,146 We next sought to determine the extent to which well-defined proteasome inhibitors might block LPS-induced inflammation.

To address this question, we pretreated RAW 264.7 macrophages with the well-characterized proteasome inhibitor, lactacystin, and observed a dose-dependent inhibition of LPS-induced cytokine secretion.^145,146 Furthermore, we found that pretreatment of primary murine macrophages with lactacystin inhibited the expression of a spectrum of LPS-inducible genes, including IL-1β, IL-6, IL-12 p40 and p35, COX-2, and iNOS. In addition, lactacystin also blocked the LPS-induced upregulation of TLR2 mRNA, and reduced constitutive levels of TLR4 mRNA expression.¹⁴⁵ The net effect of proteasome activation would appear to be enhancement of TLR-mediated inflammatory responses, while proteasome inhibition would be predicted to suppress the inflammatory response. Our data demonstrate that more than 90% of LPS-responsive genes in peritoneal macrophages are regulated by the proteasome.¹⁴⁷ Furthermore, studies from our laboratory and others suggest that the proteasome regulates a number of proteins involved in tolerance, including SOCS-1, SOCS-3, IRAK-M, IRAK-1, MyD88, TLR4, and others (Fig. 2).^147,148 In addition, the proteasome also regulates NFκB, a critical transcription factor for many LPS-responsive genes that has been shown to be dysregulated in LPS-tolerant cells. The role of the proteasome in tolerance remains largely untested thus far, however.

Fig. 2.

Schematic diagram of tolerance-related mediators that are regulated by the proteasome at either the transcriptional or post-transcriptional levels.

Mechanisms of tolerance of other lipid A structures and LPS antagonists

In addition to lipid A moieties with agonist activity, there also exist a variety of lipid A analogs that that can function as LPS antagonists. The mechanism of the LPS antagonists is likely through the competitive inhibition of LPS binding to either LPS binding molecules, such as LPS binding protein (LBP), or the TLR complex. Indeed, evidence for this has been presented for RsDPLA, the biologically inactive lipid A molecule from Rhodobacter sphaeroides, using colloidal gold particles to label both LPS and RsDPLA and electron microscopy to monitor cellular binding and internalization. Our studies conducted thus far suggest that RsDPLA competes with LPS in binding to LBP, CD14, and TLR4.^149-151 Other lipid A analogs, including ONO-4007 and others described earlier, have been postulated to induce tolerance to LPS and lipid A through upregulation of endogenous corticosteroids.⁶⁹ In addition, the inhibition of suppressor T cell activity by MPL and RsDPLA has also been proposed as a mechanism of tolerance for these two lipid A molecules.^152-154

Future directions

As summarized in this review, evidence has been presented to suggest that multiple LPS/lipid A-induced signal transduction intermediates and other mediators are involved in the induction of tolerance. It is likely that no single mechanism will emerge as playing the dominant role in this process. The continued investigation of the numerous factors implicated thus far in the development of tolerance to LPS/lipid A will help to determine to what extent and under which circumstances these various factors play a role in this phenomenon. In addition to the aforementioned factors that are currently being investigated, other factors, such as the proteasome that have not yet been widely studied in tolerance, may also be involved. Studies from this laboratory strongly suggest a key role for the proteasome in LPS/lipid A signaling. This evidence includes the modulation of proteasomal protease activity by LPS/lipid A, the degradation of IκB by the proteasome, and the subsequent activation of NFκB that ultimately upregulates inflammatory mediators. In addition to that, proteasomal proteases have also been shown to degrade various mediators of LPS signaling, including TLR4, IRAK-1, etc. Furthermore, inhibition of the proteasome modulates the gene expression of many mediators of LPS signaling, including many associated with tolerance (Table 2). The proteasome may also play a key role in tolerance such that when LPS-induced inflammatory mediators increase to a certain level, compensatory mechanisms are induced to trigger the development of tolerance, possibly by modulating the activities of individual proteasomal proteases.

Table 2.

LPS-modulated, proteasome-dependent cancer related genes. LPS-modulated, proteasome-dependent genes. Thioglycollate-elicited murine macrophages were treated with the compounds, LPS and/or lactacystin.

	LPS	LPS/LACT	LACT	DESCRIPTION
SOCS3	256.51	-4.02	-3.49	Suppressor of cytokine signaling 3
SOCS1	52.35	12.87	2.34	Suppressor of cytokine signaling 1
TLR2	10.32	6.82	-1.74	Toll-like receptor 2
NFKB2	10.20	3.60	1.23	Nuclear factor of kappa light gene enhancer in B cells
STAT2	9.17	6.68	-1.57	Signal transducer and activator of transcription 2
STAT1	6.77	4.84	-1.48	Signal transducer and activator of transcription 1
TLR3	5.74	2.77	-6.63	Toll-like receptor 3
MAP3K8	4.36	1.86	-3.09	Mitogen-activated protein kinase kinase kinase 8
IRAKM	3.46	1.66	-1.38	Interleukin-1 receptor associated kinase 3
TAK1	3.32	-3.09	2.18	Mitogen-activated protein kinase kinase kinase 7
TLR1	2.83	-1.69	-1.45	Toll-like receptor 1
MyD88	2.50	1.51	-1.08	Myeloid differentiation pathway primary response gene
CD14	2.24	-1.22	-2.31	CD14 antigen
TLR4	-2.83	-6.47	-3.89	Toll-like receptor 4
TRAF6	-5.56	-2.33	2.28	TNF-receptor associated factor 6

The gene expression values are reported as average normalization ratios (modified from ref. 165). A data set containing gene identifiers and their corresponding expression values was uploaded as an Excel spreadsheet using the template provided in the application. Each gene identifier was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge Base.

Interestingly, there is a great deal of overlap between many of the proposed mechanisms of tolerance and putative targets of cancer therapy. The roles of NFκB, IRAK-1, IRAK-M, SOCS-1, SOCS-3, and MyD88 have been evaluated both in tolerance as well as in several models of cancer, for example. Because proteasome inhibitors modulate gene expression of these signal transduction intermediates, they have also been investigated for their potential as cancer therapeutics. Indeed, the proteasome inhibitor, Bortezomib has recently been approved by the FDA for the treatment of refractory multiple myeloma and has also shown promise in the treatment of lung cancer, as well as various types of lymphoma.

As our knowledge of lipid A and its derivatives continues to expand, the therapeutic potential of these compounds has become evident. MPL has been used as an effective adjuvant for cancer vaccines and will likely continue to be used in future vaccine formulations. The development of nontoxic lipid A analogs, such as SDZ MRL 953, that induce tolerance to LPS and thereby protect susceptible patients against endotoxin shock also shows therapeutic promise. Moreover, the development of lipid A analogs that induce less tolerance than LPS and exhibit greater efficacy in tumor regression show potential as cancer therapeutics as well. Because lipid A-mediated tolerance is a particularly complicated phenomenon that plays dual and opposite roles in the efficacy of cancer therapeutics, elucidation of the mechanisms of tolerance is essential for the continued development of lipid A compounds into nontoxic, efficacious treatments for cancer patients.