The Immune Context in Microbiological Testing

Abstract

The control of biologics microbiological impurities, contaminants, and mimetics is evolving. Much of the ongoing change is associated with the realization that the stimulation of the adaptive immune system is a separate but also overlapping concern to that associated with the historical preclusion of microbial artifacts modeled singularly around the innate immune response.

In microbiological control, models of contamination have historically been based upon the innate, proinflammatory responses and it has not been intuitive that artifacts generating immunostimulative responses could proceed via different sources. Beta-glucans for instance produce no overt proinflammatory provocation but have come to be known as immune modulators. The innate immune system has been found to inform adaptive responses and the transfer of information involves sophisticated cross-talk (1) with both co-stimulatory and co-inhibitory signals from Toll-like receptors (TLR) informing antigen receptors and continues to be elaborated. 

The mammalian host response to a wide range of microbial artifacts underlies the need to preclude or remove them from biologic drug manufacturing processes. Typically, host expression system remnants are called impurities, and those foreign to the expression system are called contaminants. The mammalian responses to microbial artifacts is two-pronged: an overt, initial, and rapid proinflammatory response is activated directly by microbial-associated molecular patterns (MAMPs) via pattern recognition receptors (PRRs) including TLRs, resulting in cytokine production and complement activation, followed by fever and coagulopathy, and may proceed to shock or sepsis. A second, adaptive response, if it proceeds, does so after a delay and in response to immunostimulatory prompts from the innate immune system. Here, microbial artifact MAMPs contribute to the activation of the adaptive immune system via production of different sets of cytokines (than those producing the inflammatory response), ultimately resulting in a highly specific, long-lasting antibody production by B lymphocytes. Higher organisms (beginning in Gnathesomes or jawed vertebrates) have parallel acting immune system architecture (see Figure 1).

Figure 1. The two major interfaces for microbial recognition in the innate and adaptive immune systems respectively. TLR4 is likely unique in the complexity and variety of the types of a single microbial- associated molecular patterns (MAMP) recognized (LPS). TLR is Toll-like receptor. (All figures courtesy of the author)

Control of biologics impurities, contaminants, and mimetics

Testing for bioburden, sterility, particulates (as microbial mimetics), and endotoxin for small-molecule drugs (SMDs) and large-volume parenterals (LVPs) has been well established for decades, with no need for additional control. The advent of modern biologics, however, has brought with it adverse drug reactions (ADRs) arising from drug mechanism of action alone or from impurity/contaminant interaction with immune cell PRRs and surface signaling molecules (CSSMs) to induce the generation of anti-drug antibodies (ADAs) or neutralizing drug antibodies (NDAs). The rapid rise of immunological knowledge and the realization by some (including some at FDA labs: Verthelyi and Wang [2] and Haile et al. [3]) that microbial impurities/contaminants/mimetics should be gauged differently when dealing with biologic drugs is bringing change to microbiological analytical testing. 

The move from a singular focus on microbiology to a broader view of microbiology with an associated immune context is a change from the detection of overt to covert contaminants. Immunogenicity concerns dominate biologics drug development and have largely been handled upstream by drug developers using sophisticated tools including “humanization” of the molecules (4) and drug immunogenicity testing in mice (even “humanized” mice [5]). The layering onto existing tests of an immune context has been viewed by some an additional burden to microbiological analytics. This can be seen in efforts to develop host cell protein tests (6, 7), the use of polymerase chain reaction (PCR) for nucleic acid detection (8), the detection of subvisible particles as well as the addition of “contaminants” not previously viewed as such. The difficulties associated with increasing expectations are apparent: “it is quite a challenge to detect and measure host cell proteins (HCPs) in a matrix dominated by the recombinant protein” (9). Similarly, the addition of low endotoxin recovery (LER) and endotoxin-protein binding concerns constitute an additional class of “hidden endotoxin” to analytic detection efforts. Examples of biologics assays that have testing associated with an immune-context include: HCPs (10), nucleic acids, β-glucans (11, 12), and endotoxin (13). Perhaps the first indication that biologics control would be different occurred in the recognition that microbial mimetics can bring unwanted attention (ADAs, NDAs) to biologic proteins and includes: sub-visible particles (SVPs), protein aggregates, and emulsions.

Examples of adverse events that have precipitated the addition of immune context are plentiful from microbial-derived and microbial mimetic perspectives, including: 
 

• Flagellin as a HCP impurity co-purified with drug substance (14). 

• Viral RNA/DNA as well as mycoplasmas have been found in some biologics manufacturing processes (15). 

• β-glucans, once viewed as harmless fungal-derived sugars, are now viewed as immune reactive (16).

• Subvisible particles in Omontys, a synthetic erythropoietin peptide mimetic drug, were identified as a potential cause of anaphylaxis and deaths (17).

• Protein aggregation exacerbated by silicone oil (emulsions) has been seen from stoppers and vials (18).

If some assay analytes (including masked endotoxin) have not been implicated by precipitating events, it is likely due to the complexity of making such an association between a given adverse reaction and its causative agent. 

The visibility of adverse events is hampered by the mode of action of various biologics that are prone to induce fever/pyrexia and infusion reactions as a common administration side effect. Problems arising from fever-causation due to impurities and contaminants are believed by many to have already been solved and, therefore, all reactions are thought to be associated with the drug itself or attributed to unknown causes. Baldo relates the “state of the art” of biologics adverse reactions in Safety of Biologic Therapy (19):

“Infusion of many biologics, particularly monoclonal antibodies (mAbs), provokes a characteristic infusion syndrome, usually within one or a few hours during/after the first administration. Whereas most reactions are mild to moderate with symptoms often described as ‘flu’-like with fever, chills, rigors (shaking from high fever), headache, nausea, asthenia, rash, and pruritus, a small number of patients, mostly at the first or second infusion, show potentially fatal symptoms ... The mechanisms of mAb-induced infusion reactions are not yet fully understood” (19). 

Thus, the drug mode of action does not explain all instances of fever or drug -infusion reactions. What is interesting is the similarity of this common biologics “infusion reaction” to endotoxin infusion, historically performed in humans to establish the threshold pyrogenic response: “endotoxin doses of 2-4 ng/kg body weight cause flu-like symptoms (fever, chills, myalgia, headache, nausea)” (20). Some experts have voiced doubt as to the inherent nature of deleterious immune reactivity of mAbs (as human molecules) and other biologic drugs: 
 

• “In the absence of adjuvants, proteins are usually not immunogenic, in fact, often, they are tolerogenic. That is the basis of allergy immunotherapy…” (21).

• "The unexpected development of immune responses to fully human antibodies and proteins … has become one of the greatest puzzles of the protein therapeutics revolution” (22).

• “Despite the general trend towards ‘humanization,’ these drugs remain immunogenic in clinical settings, baffling drug developers. In principle, humanized and fully human monoclonal antibodies are ‘self’ immunoglobulins and should be tolerated” (23).

Given the historical relationship of proteins and endotoxins, the presence of hidden, low dose, or detoxified endotoxins should not be completely ruled out (to be discussed). However, fever is a minor response compared to immunogenicity, which provokes the adaptive response against the therapeutic protein and sometimes results in the cessation of life-saving therapy. Technically, expected fever is not an adverse drug reaction (ADR) because Code of Federal Regulations (CFR) 314.80 states that drug manufacturer reports required within 15 days of receipt are only required: “to report to FDA individual case safety reports for events classified as ‘unexpected’ or ‘unlabeled’ (not described in the product package insert) and ‘serious.’ Furthermore, FDA MedWatch.org reports are voluntary (24) and as such cannot provide statistical evidence of occurrence frequency. 

Immune context defined

Simply stated, the immune context for biologics microbiological contaminants is an inclusion of the detection and exclusion of “immunostimulatory artifacts” that can alert the adaptive immune system of the presence of large proteins in a negative way. Immunostimulatory artifacts could be considered “hidden” in several ways, including the lack of knowledge of what artifacts may be immune stimulating. Analytically speaking, one cannot find what one does not seek and one is likely to find only a fraction of what one does seek. A wide variety of mammalian receptors (10 TLRs in man) cover dozens of microbial artifacts including the lipopolysaccharide (LPS) receptor (TLR4). Viewing endotoxin as a model MAMP, it has both proinflammatory and immunostimulatory properties. These proceed by different pathways after receptor dimer complex (TLR4/MD2) activation. One is called the MyD88-dependent pathway and the other is the MyD88 independent pathway (25,26). The detoxification of endotoxin for vaccine research demonstrates the dual nature of LPS. Historically, detoxification has been achieved by many different LPS treatment methods and effectively severs the toxic (proinflammatory) response to LPS from its immunostimulatory properties (25). For a list of detoxification methods, see reference 27.

The innate response has been found to be critical in terms of informing the adaptive response (28). Janeway (29), whom first proposed the very idea of a PRR in 1989, predicted that highly conserved innate PRRs (including TLRs) contain the information necessary to identify and respond to MAMPs like LPS. In contrast, the adaptive immune system, given the hyper variability-generating systems used to make a million different antibody structures by shuffling small sets of germ-line encoded genes is literally guessing at contaminant structures. Though there are some complex methods of weeding out self-reactive versions, the largely random guesswork is so prolific that it is all but certain to bind (via paratope) any given protein target (epitope). Thus, the adaptive structures, membrane-bound antibody receptors on B-lymphocytes for example, need the context provided by TLRs (the most potent activator thereof being LPS), otherwise deleterious autoimmune binding events would be much more frequent: 

“Because the gene shuffling mechanisms used to generate diversity among T cell receptors (TCRs) and B cell receptors (BCRs) can end up recognizing self, the cells of the adaptive immune system require instruction by the cells of the innate system as to whether an immune response should be mounted to a particular antigen (or not). This is a critical role …” (30). 

And

“… the semantic information is conveyed by the non-clonal recognition system (PRRs), because randomly created receptors cannot carry semantic content as they would not know in advance what antigen they will recognize and what type of response they will have to induce” (31).

Figure 2 shows a basic difference in microbial interfacing between the largely randomly generated antibody receptors and the genetically evolved and conserved PRRs such as TLRs. 

Figure 2. Mammalian-microbial interfaces of adaptive and innate immune responses via antigen receptor and TLRs, respectively. Based upon reference 1 (Figures 1 and 2). Given that mammalian paratopes are a product of random rearrangement, conserved PRRs (including TLRs) inform the adaptive response. The phenomenon is also the concept of vaccine adjuvant behavior as referenced by FDA Immunogenicity Guideline (2014). BCR= B-cell receptor, TLR = Toll-like receptor, MAMPs are microbial associated molecular patterns such as LPS, PRR = pattern recognition receptor.

FDA developed an immune context for microbiological testing but the idea remains obscure to many in the industry. The FDA guidance document, Immunogenicity Assessment for Therapeutic Protein Products, section 5, page 18, describes “Impurities with Adjuvant Activity” and the activation of the adaptive immune system by the innate immune system (32).

Difficulties of expanding control

The historical views of adjuvanticity (purposeful contamination of a protein for vaccination purposes) line up with biologics contaminant control efforts in that the types of impurities/contaminants encountered represent historical adjuvant classes: (a) particulates and aggregates (aluminum salts), (b) emulsions (mineral oil, squalene), and (c) microbial derived MAMPs (mycobacteria cell walls, LPS, etc.). The LPS, monophosphoryl lipid A (MPLA) is FDA approved as a vaccine adjuvant (33). The prototypical Escherichia coli (E. coli) LPS (hexa-acyl, biphosphoryl) has both proinflammatory and immunostimulatory capability but through detoxification (acid-base hydrolysis), MPLA is produced, which is minus one acyl and one phosphate group. This detoxified LPS is 1000 times less proinflammatory but remains 10-20 times more immunostimulatory than a protein without an adjuvant (34). There are also natural forms of LPS that are not proinflammatory, which could, eventually, have profound implications for control (35). MPLA was “The first TLR agonist to be used intentionally and expressly for its immunostimulatory properties” (36). 

Figure 3 shows a simplified view of the complex interactions of LPS with TLR4. It is complicated because small structural changes can bring vast signaling changes. In proposing new classes of immune-reactive substances, “hidden endotoxin” would certainly qualify, indeed provides a prototypical class. It is immunostimulatory either at low doses, after detoxification (27, 33, 37) or as masked by protein or surfactant. Some natural forms that are either non-or low pyrogenic have also been shown to have immune-stimulating properties (35). This makes the detection of masked or hidden endotoxin an important practical -analytical avenue to advance biologics drug control capabilities. 

Figure 3. Prototypical E. coli hexa-acyl LPS optimizes dimer receptor complex to produce proinflammatory responses in mammals, whereas other structures result in less potent or adaptive immune stimulatory responses. Tetra-acyl drug candidate Eritoran has the glucosamine backbone rotated 180 degrees as it sits in MD-2, blocking receptor activity. (Information sourced from references 37, 41, 42).

LER arises from specific drug formulations that disassociate LPS aggregates into monomers and multimers via chelation (citrate, phosphate, etc.) followed by polysorbate masking (the small surfactant molecules are typically in abundance). The phenomenon thus blocks the ability of Limulus-based assays to detect endotoxin; however, in activating cellular surface markers and adaptive immunity, it cannot be assumed that masked endotoxin is inert rather than immunostimulative (even if proinflammatory activity is missing or muted), and indeed this theoretical immune reactivity has recently been demonstrated (38). LER and protein masking can be overcome by demasking, however, demasking remains developmentally challenging. 

Of the 50 approved mAb drugs listed by Baldo (approved mAbs as of June 2016 excluding biosimilars), 41 have an LER type of formulation. Therefore, it’s an important issue given the widespread use of such formulations. Polysorbate is used to prevent another important immune-context-related issue: protein aggregation. Given that LER is not the new paradigm but rather an example of the paradigm (immunostimulative artifacts), more types of relevant masking and/or detoxification will likely be discovered. Indeed, one new type includes the use of detergent upstream (Triton X-100) in protein extraction that can bring LER downstream without polysorbate (39). 

The tenacity of endotoxin-protein binding has been known for many years. Protein-endotoxin binding in vaccine proteins, either knowingly (as adjuvant) or unwittingly (as contaminant) has been extensive. See reference (39), Table 1 (“Reported endotoxin levels for various vaccines”), for a list of endotoxin measurements in historical vaccines. The Petsch study (40) demonstrated the complex relationship of protein and endotoxin. Petsch showed a large amount of endotoxin-protein binding even at low protein concentration and neutral pH (1 mg/mL / pH = 7). Indeed, most biologic drug product concentrations are greater than 1 mg/mL and many are much greater (more protein brings more binding capacity). Cationic proteins (positively charged) bind endotoxin which has a strong negative charge. The pI is the pH where protein charge is neutral, thus a high pI will be highly positive in pH neutral solutions. A table of the Petsch study findings are shown in Table I. Note the tenacity of IgG-endotoxin binding, even after destructive testing with a protease. This is important as IgG is the scaffold for monoclonal drug proteins.

Conclusion

The control of biologics microbiological impurities, contaminants, and mimetics is evolving. Much of the ongoing change is associated with the realization that the -stimulation of the adaptive immune system is a separate but also overlapping concern to that associated with the historical preclusion of microbial artifacts modeled singularly around the innate immune response. Limulus has no adaptive immune system, for example, and a rabbit pyrogen test is a three-hour test that does not detect antibody formation. This distinction is important to biologics manufacturing due to the propensity of ADA and NDA formation against protein therapeutics. Limulus-based tests remain the most sensitive and practical tests for endotoxin by far, but sample treatments to reveal masked endotoxin can be more widely explored in process development and routine control.

Endotoxin is viewed as a model contaminant as it is the most potent, ubiquitous, and stable microbial artifact. Control has focused on its proinflammatory properties, including fever. It is less appreciated in pharmaceutical control (as opposed to vaccinology) that it is also highly immune stimulatory and it can do so even at low doses or after the proinflammatory response has been muted by various treatments (detoxification). Endotoxin that is protein or surfactant bound in a production process may not be pyrogenic or LAL reactive, yet this is not a guarantee that it is not immune stimulating. Indeed, biologics microbiological control is based upon theory and now the overarching theory is being revised. This paradigm change will continue to be debated; however, it is being enacted at the same time and proceeds in a step-wise fashion to cover various forms of adaptive immune system activating molecules as various correlations are established. 

References 

1. Zhu et al., Immunity, 34(4): 466-478 (April 22, 2011).
2. Verthyli and Wang, PLoS ONE, 5(12): e15252 (December 2010). 
3. Haile et al., PLoS ONE 10 (4): e0125078, (April 2015).
4. Getts et al., mAbs 2 (6), 682-694 (November/December 2010).
5. Schultz et al., Nature Reviews Immunology, 7, 118-130 (February 2007). 
6. Bracewell et al., Biotechnology and Bioengineering, 112 (9), pg 1712-1737 (September 2015). 
7. Bomans et al., PLoS ONE, 8 (11), pg 1-11 (November 2013). 
8. Hyuck et al., J. Microbiol. Biotechnol., 20 (10), 1463-1470 (2010).
9. Wang, Richardson, and Shameem, BioPharm International, 28 (6), pg 32-38 (June 2015).
10. Wang, Hunter, and Mozier, Biotechnology and Bioengineering, 103 (3) (June 15, 2009).
11. Gefroh et al., Biotechnol. Prog. 29 (3) 672-80 (May-June 2013).
12. Vigor et al., Biotechnology Process 29(3):672-80. (May-Jun 2013). 
13. Hughes et al., “Low Endotoxin Recovery: An FDA Perspective,” BioPharm Asia, April 10, 2015, accessed June 8, 2017,  https://biopharma-asia.com/magazine-articles/low-endotoxin-recovery-an-fda-perspective/
14. Huang et al., Jour. Clin. Microbiology, 47(11): 3427-3434, Nov. 2009.
15. Geigert, “Demystifying CMC Regulatory Strategy for Biologics Part 4: Challenges of Adventitious Agent Control,” RegulatoryFocus.org, March 2012.
16. John A. Bohn and J. N. BeMiller, Carbohydrate Polyners 28, 3-14 (1995).
17. Kotarek et al., Journ. Pharm Sci. 105, 1023-1027 (2016).
18. Carpenter et al., Jour. Pharm Sci. 98(9): 3167-3181, (Sept. 2009). 
19. Brian Baldo, Safety of Biologic Therapy (Springer, 2016).
20. DellaGioia, and Hannestad, Neurosci Biobehav Rev. 34(1): 130-143 (January 2010).
21. P. Lollar, “The Immune Response to Blood Coagulation Factor VIII,” 14th Annual Immunogenicity for Biotherapies, Baltimore MD, March 2013, Institute for International Research, New York, N.Y.
22. Scott and Groot, Ann Rheum Dis, 69 (Suppl I): i72-i76 (2010).
23. De Groot et  al., Expert Rev. Clin. Pharmacol. 6(6), 651-662 (2013).
24. FDA MedWatch, www.fda.gov/safety/medwatch/ 
25. Zughaier et al., Infection and Immunity, pg. 2940-2950, (May 2005).
26. Wang et al., Biol Rev Camb Philos Soc. 90(2):408-27, (May 2015).
27. K. Williams, BioPharm Intl., 28 (2), pg, 24-31 (February 2016).
28. Iwasaki and Medzhitov, Nat Immunol. 16(4): 343-353, (April 2015).
29. Janeway, Pillars of Immunology 54: 1-13 (1989).
30. Delves et al., Roitt’s Essential Immunology (Wiley, 2017).
31. Fierz, Frontiers in Immunology 7 (551) (December 2016).
32. FDA, Immunogenicity Assessment for Therapeutic Protein Products, 2014. 
33. Vandepapeli`et al., Vaccine 26, 1375–1386, (2008). 
34. Casella and Mitchell, Cell Mol Life Sci.65(20): 3231-3240, Oct. 2008.
35. Chilton et al., Frontiers in Immunology 3 (154) (June 2012).
36. Bohannon et al., Shock, 40(6): 451-462, (Dec. 2013).
37. L. Bertok, Pathophysiology 12, 85-95 (2005).
38. Schwarz et al., Nature, Scientific Reports, 7 : 44750, pg 1-11, (March 2017).
39. L.A. Brito and M. Singh, Jour. Phar Sci. 100 (1) (January 2011).
40. Petsch D, Anal Biochem. 15;259(1):42-7, (May 1998).
41. Kim et al., Cell, 130 (5):906-17 (Sept. 7, 2007). 
42. Chilton et al., Frontiers in Immunology, 3, article 154 (June 2012). 

Article Details

BioPharm International
Volume 30, Number 9
September 2017
Pages: 32–38

Citation

When referring to this article, please cite it as K Williams, “The Immune Context in Microbiological Testing," BioPharm International 30 (9) 2017.