23 Feb 2023: Smith DFQ, Casadevall A (2023) Correction: On the relationship between Pathogenic Potential and Infective Inoculum. PLOS Pathogens 19(2): e1011191. https://doi.org/10.1371/journal.ppat.1011191 View correction
Pathogenic Potential (PP) is a mathematical description of an individual microbe, virus, or parasite’s ability to cause disease in a host, given the variables of inoculum, signs of disease, mortality, and in some instances, median survival time of the host. We investigated the relationship between pathogenic potential (PP) and infective inoculum (I) using two pathogenic fungi in the wax moth Galleria mellonella with mortality as the relevant outcome. Our analysis for C. neoformans infection revealed negative exponential relationship between PP and I. Plotting the log(I) versus the Fraction of animals with signs or symptoms (Fs) over median host survival time (T) revealed a linear relationship, with a slope that varied between the different fungi studied and a y-intercept corresponding to the inoculum that produced no signs of disease. The I vs Fs/T slope provided a measure of the pathogenicity of each microbial species, which we call the pathogenicity constant or kPath. The kPath provides a new parameter to quantitatively compare the relative virulence and pathogenicity of microbial species for a given host. In addition, we investigated the PP and Fs/T from values found in preexisting literature. Overall, the relationship between Fs/T and PP versus inoculum varied among microbial species and extrapolation to zero signs of disease allowed the calculation of the lowest pathogenic inoculum (LPI) of a microbe. Microbes tended to fall into two groups: those with positive linear relationships between PP and Fs/T vs I, and those that had a negative exponential PP vs I relationship with a positive logarithmic Fs/T vs I relationship. The microbes with linear relationships tended to be bacteria, whereas the exponential-based relationships tended to be fungi or higher order eukaryotes. Differences in the type and sign of the PP vs I and Fs/T vs I relationships for pathogenic microbes suggest fundamental differences in host-microbe interactions leading to disease.
The ability of a microbe, virus, or parasite to cause disease is dependent on multiple factors, virulence factors. host immune defenses, the infective inoculum, and the type of immune response. For many microbes their capacity for causing disease is highly dependent on the inoculum. The mathematical formula for Pathogenic Potential (PP) is a way to compare the ability of an organism to have a pathogenic effect, as measured by Fraction with signs or symptoms (Fs), mortality (M), and inoculum (I), and can include the median survival time of the host (T). Increasing inoculum of the fungus Cryptococcus neoformans for a moth host resulted in exponentially smaller pathogenic potential, and the Fs/T versus inoculum plot showed a logarithmic relationship. Together, these relationships show diminishing returns with increasing cryptococcal inoculum, in which each individual fungus plays a smaller role in pathogenicity. Literature data shows that other microbes, mostly bacteria, had linear Fs/T versus inoculum relationships, which indicate that each bacterium contributed an equal amount to pathogenicity. These differences in relationships can point to differences in host-microbe interactions and suggest new ways in which the organism causes disease.
Citation: Smith DFQ, Casadevall A (2022) On the relationship between Pathogenic Potential and Infective Inoculum. PLoS Pathog 18(6): e1010484. https://doi.org/10.1371/journal.ppat.1010484
Editor: Michal A. Olszewski, University of Michigan Health System, UNITED STATES
Received: March 30, 2022; Accepted: May 16, 2022; Published: June 13, 2022
Copyright: © 2022 Smith, Casadevall. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: D.F.Q.S. and A.C. were supported in part by National Institutes of Health (https://www.nih.gov) grants R01-AI162381, R01-AI152078, and R01-HL059842. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The pathogenic potential (PP) of an organism was proposed in 2017 as an attempt to develop a quantitative method that would allow comparing the capacity for virulence of different microbial species  and is defined by the equation: Eq 1 whereby Fs is the fraction of the population with microbe-relevant signs or symptoms, I is the infective inoculum, and M is the mortality fraction. Mortality, as the 10M term, was included as a separate variable in order to amplify the pathogenic potential of lethal microbes versus non-lethal ones. In host-microbe interactions that do not result in host death that M = 0.0 and the term 10M becomes 1. Later this concept was expanded by showing how PP could be used to estimate the contribution of virulence factors to pathogenicity, and by adding the parameter of time, described as PPT (Eq 2), to account for the fulminant nature of some infectious diseases .Eq 2
The initial equations were written assuming that the various parameters were linearly related as a first approximation, partly for simplicity and partly because there was no evidence to the contrary. However, proposing a PP equation raised the question of what the actual mathematical relationship between such parameters as Fs and I was, which in turn suggested the need for experimental measurements using pathogenic organisms in a susceptible host. A further question was whether there were differences between these parameters in different microbial species or hosts. For example, vertebrates have both innate and adaptive immune responses that neutralize microbes, whereas invertebrates have only an innate-like immune response. Further, the mechanisms by which microbes damage hosts and cause disease vary widely. Disease occurs when the host has suffered sufficient damage such that homeostasis is altered and this damage can come from direct microbial action, the immune response, or both .While each pathogenic microbe is different and generalizations are difficult, bacteria tend to cause disease through routes of tissue damage and toxicity, whereas fungi cause disease through growth in tissues and persistence within the host, and for both host damage results from microbial action and the immune response. Consequently, we hypothesized that differences in mechanism of disease could be reflected in differences in the relationships between the measures of pathogenicity and inoculum.
In this study we used the Galleria mellonella system  to explore the relationship between I and Fs. This system is particularly attractive because it is a non-vertebrate animal host that is highly susceptible to many pathogenic microbes. Our analysis reveals a non-linear relationship between PP and I and suggest that the slope of the relationship between I and Fs can be used for a quantitative comparison of the relative virulence of microbial species. We investigated the existing literature to evaluate whether this exponential relationship between PP and I was universal or unique to C. neoformans in the G. mellonella host, and found that other microbes, predominately bacteria, had linear PP vs I and Fs/T relationships, whereas fungi tended to have the exponential relationships seen with C. neoformans. Further, we see the same exponential relationships with C. neoformans infections of murine hosts as we do in G. mellonella hosts. Our results suggest that the types of mathematical relationships can differ for individual pathogenic microbes and that these differences can reveal fundamental differences in virulence strategies and/or host responses to infection.
Pathogenic Potential for Cryptococcus neoformans in Galleria mellonella
We analyzed the pathogenic potential (PP) of C. neoformans H99 strain when infecting Galleria mellonella at an inoculum of 105 cells/larvae and incubated at 30°C from sixteen different experiments (Fig 1A) and found that the average PP was 8.64 x 10−5 (Fig 1B). We similarly calculated the PPT, which is a measurement of pathogenic potential as it relates to time until death in 50% of hosts (LC50). We found that the average PPT of C. neoformans at this inoculum was 1.23 x 10−5 (Fig 1B). These data from 16 independent experiments shows the variation in PP measured in one laboratory and provides a range to consider when comparing calculated PP and PPT from other organisms using literature values below.
(A). Overlapping plots of the survival of G. mellonella infected with C. neoformans at an inoculum of 105 cells/larvae. Each of the 16 survival curves represents a replicate infection with 15 to 30 larvae. The red line indicates the combined survival curve with a 95% confidence interval. The individual pathogenic potential (PP) (B) and pathogenic potential in respect to time (PPT) were calculated and plotted. Each data point in (B) represents the calculated PP or PPT of an individual experiment. Error bars represent mean with 95% confidence interval.
Correlation of PP and PPT as a Function of Inoculum
To understand the relationship between inoculum and PP and PPT, we infected G. mellonella with C. neoformans using different inoculums (Fig 2A). We observed that as inoculum increased, there was an expected decrease in time to death until 50% of host organisms died, with an increase in Fraction with signs or symptoms (Fs) and Mortality (M) (Fig 2B). We also observed a negative exponential decrease in PP and PPT while inoculum increases (Fig 2C and 2D). In both measures, the lower inoculum was associated with a higher pathogenic potential. The negative exponential relationship between pathogenic potential and inoculum implies that the average microbe during an infection with high inoculum makes smaller contribution to the outcome of infection than in a lower inoculum infection.
(A) Survival curves of G. mellonella infected with different inocula of C. neoformans, and the calculated Fs, T, M, and pathogenic potentials (PP and PPT) (B). Plots of PP (C) and PPT (D) versus I on log-scaled x-axes. These show a negative exponential relationship between pathogenic potential and inoculum, as fitted by a one phase exponential decay function. 95% CI of the exponential fit line is shown with dotted lines.
Correlation of Fs/T as a Function of Inoculum
Plotting Fs versus Inoculum yielded logarithmic curves (Fig 3A). Similarly, a plot of Fs/T versus I revealed a logarithmic relationship (Fig 3B). The higher the inoculum, the higher the Fs and Fs/T values are. Further, the relationship between Fs/T and the log of the inoculum was linear, indicating a direct correlation between log(I) and Fs/T (Fig 3C), implying that a simple line equation described that relationship. Since this relationship between inoculum and disease is logarithmic and not linear, it implies that microbes the higher inoculum on average contribute less to the outcome of infection. This would be consistent with a microbe that causes disease from a high microbial burden in the host due to exponential doubling of microbes. From this line equation, we could derive the y-intercept, which would be the smallest inoculum to cause a pathogenic effect with regards to time (Fs/T), which we termed the Lowest Pathogenic Inoculum (LPI) (Fig 3C). Similarly, the slope provided information on how initial inoculum is related to the outcome of the host, and by virtue of being a slope is a constant value that describes the microbe’s pathogenic nature regardless of inoculum.
(A) Fraction with signs or symptoms and (B) Fraction with signs or symptoms relative to the LT50 for larvae infected with different inocula of C. neoformans plotted on a log-scaled x-axis. These show that there is a positive logarithmic relationship between Fs and Fs/T versus inoculum, as fitted by a semi-log line, in which the x-axis is logarithmic, with 95% CI shown as dotted lines (A and B), or a simple linear regression for log(I) vs. Fs/T (C). This relationship can be used to calculate the pathogenicity constant (kPath) and the lowest pathogenic inoculum (LPI) (C).
Similarities between PP, PPT, and Fs/T in Mice and G. mellonella models
Calculating PP and PPT for H99 murine infections showed similar trends as the G. mellonella data (Fig 4A and 4B), with both having negative exponential relationships between the measures of pathogenicity and the inoculum of infection, for the different mouse strains and route of infection. This suggests similar relationships between the host and C. neoformans in both G. mellonella and murine models. When calculating Fs/T values from H99 murine infections, we found similar trends in the Fs/T values, indicating similar relationships between the host and C. neoformans in both G. mellonella and murine models (Fig 4C and 4D). The data also indicated lowest pathogenic inoculums (LPI) that varied by mouse strain and route of infection, some of which were comparable to the LPI of C. neoformans in G. mellonella. For intravenously infected C57BL/6 mice, the LPI was 14.7 cells while for intranasal infection the LPI was 4830 cells. The intravenous-infected C57BL/6 also had a lower LPI than the intravenous-infected ICR strain (288 cells), which could be indicative of immune variations between the strains.
Using literature values [47–49], we calculated the (A) pathogenic potential (PP), (B) pathogenic potential in respect to time (PPT), (C) Fs/T, (C) lowest pathogenic inoculum (LPI), and (C) kPath for C. neoformans in mouse models through various inoculation routes. Generally, the trends were consistent between the fungus in G. mellonella and murine hosts. (A) PP vs I and (B) PPT vs I data was fitted by a one phase exponential decay function, (C) log(I) vs. Fs/T was fitted by a linear regression, and (D) Fs/T vs I data was fitted by a semi-log line in which the x-axis is logarithmic. The (C) log(I) vs. Fs/T and (D) Fs/T vs I slopes were similar between the two hosts, indicating similar kPath values.
Pathogenicity Constant for C. neoformans
We observed a linear relationship between Fs/T and log(I) and noted that the slope of this linear best-fit equation incorporated all the components of pathogenicity (i.e fraction with signs of disease, median time until death (LT50), and inoculum into a value that is constant at all inoculums. This constant value (slope) could allow comparisons between microbial strains and species even between experiments performed at different inoculum, which is where the PP and PPT values have their limitations. Using the equation of the line derived from Fig 3C, this pathogenicity constant, kPath, can be described by (Eq 3.1–3.2).Eq 3.1Eq 3.2
The calculated value of kPath for C. neoformans (H99) infection of G. mellonella is 0.0369 based on our experimental data. We calculated a kPath for C. neoformans infections in mice ranging from 0.032 to 0.046 depending on the mouse strain, route of infection, and study (Fig 4C), which is comparable in magnitude to that for G. mellonella. The kPath value is defined as the fraction of hosts with signs of disease per LT50 log inoculum. Essentially, kPath is a measure of how fast the hosts get sick and die per log inoculum. High values represent microbes that cause greater and faster damage with each additional order of magnitude of cells, conversely, smaller values represent microbes that cause a steady, slower pathogenicity in which additional orders of magnitude of cells do not have a substantial effect.
Fungal PP, PPT, Fs/T and kPath in G. mellonella
From these insights with the C. neoformans-G. mellonella system we explored their applicability to other pathogenic microbes and analyzed published G. mellonella data to calculate the experimental PP, PPT, Fs/T and kPath of other fungi. For the entomopathogenic fungus Beauveria bassiana, the relationships between fungal inoculum and PP, PPT, and Fs/T were each similar to those calculated for C. neoformans with a slightly higher kPath equal to 0.1 (Fig 5A, 5B and 5C) [5,6]. However, we saw different trends for the three other fungal species. In the case of Candida albicans, there was no clear relationship between inoculum and PP and PPT, however, the Fs/T versus I relationship was logarithmic, like B. bassiana and C. neoformans, but with a much steeper slope, and thus the higher kPath of 0.566. (Fig 5D, 5E and 5F, black) [7,8]. Similar trends and values were seen in G. mellonella infections performed by our group (Fig 5D, 5E and 5F, teal). The steeper kPath and the higher LPI indicate there is a higher barrier for the fungus to be pathogenic, but once that threshold is met, pathogenicity increases rapidly. For G. mellonella infected with Histoplasma capsulatum and Paracoccidioides lutzii, the plotting yielded negative exponential relationships between inoculum and PP and PPT, and an Fs/T vs I relationship that was essentially flat with a kPath value near zero (Fig 5G, 5H and 5I) . Essentially, based on the Fs/T vs I and kPath values, there was no inoculum-dependent mortality for the infected larvae for these two pathogenic fungi. However, in one study  that used a higher inoculum, there was a dose-dependent effect on host death, where larvae infected with 5 x 106 cells died faster than those infected with 1 x 106. Future studies may want to further investigate the mechanism underlying the unique dose-dependency, or independency, of H. capsulatum and P. lutzii infections in G. mellonella. Further investigation may include quantification of the reported dose-dependent melanization response in larvae, which could be used as the Fs value and provide more nuanced and intuitive inoculum dose-dependency in PP, PPT, and Fs/T. The general dose-independent effect on survival could be the result of the slow and irregular growth of the microbe [11–13], or a damaging immune response that kills the host in response to few or many microbes (Table 1). In this regard, P. lutzii, P. brasiliensis, and H. capsulatum are both slow growing fungi with doubling rates in media ranging from 13 to 21 hours [11–13], compared with the ~2 hour doubling time of C. neoformans in culture  and ~5 hours in vivo during infection of G. mellonella hosts . Associations between P. brasiliensis growth rate and virulence have been previously indicated . Additionally, the higher temperatures for the 37°C conditions used in these experiments is a variable that may cause thermal stress on the larvae that could impact their immune response and baseline longevity compared to the 25°C incubation condition [17,18]. Dissimilar to the findings in G. mellonella, analysis of murine infection with P. brasiliensis reveals an inoculum-dependency for PP, PPT, and Fs/T similar to what is seen with other fungi (S1A, S1B, S1C, and S1D Fig) . There is no clear inoculum-dependent effect on PP or PPT in H. capsulatum infection of mice, and there is a roughly positive linear relationship between Fs/T versus I (S1E, S1F, S1G and S1H Fig) [20,21], which is unlike other fungal Fs/T vs. I relationships observed in G. mellonella.
Using existing published values [5–10], we calculated (A) PP, (B) PPT, and (C) Fs/T for the entomopathogenic fungus Beauveria bassiana’s. These showed similar relationship to inoculum as C. neoformans. Similarly, we calculated C. albicans’ (D) PP, (E) PPT, (F) and Fs/T and plotted it versus inoculum from previously published and new experimental data. We did not see a clear association of PP and PPT with the inoculum, however, there was a logarithmic relationship between the inoculum and Fs/T (F). For Histoplasma capsulatum, Paracoccidioides lutzii, and Paracoccidioides brasiliensis, we used literature sources to calculate the (G) PP, (H) PPT, and (I) Fs/T vs. inoculum with different strains and temperatures and found that the PP and PPT mostly had a relationship with inoculum that was best fitted by a one phase exponential decay line. The Fs/T values were mostly independent of inoculum used, with the exception of the Pb18 and Pl01 strains at higher inoculums. (A,D,E) PP vs I and (B,E,H) PPT vs I data was fitted by a one phase exponential decay function, and (C,F,I) Fs/T vs data was fitted by a semi-log line in which the x-axis is logarithmic.
Bacterial PP, PPT, Fs/T in G. mellonella
Next, we considered data found in literature that would allow us to calculate PP, PPT and Fs/T for bacterial infections of G. mellonella [22–27]. In general, the relationships between pathogenicity and inoculum for bacteria were different from those relationships in fungi. For example, all the bacterial species analyzed, aside from Salmonella enterica Typhimurium had an Fs/T vs I relationship that was linear, compared to the logarithmic one in fungi (Fig 6C, 6F, 6I, 6L and 6O. This indicates a direct relationship between disease progression over time and the starting inoculum, rather than one related to the inoculum’s order of magnitude (log[I]). The positive linear relationship between Fs/T and inoculum indicates that microbes contribute equally to disease during low and high inoculum infections, meaning that each bacterium makes a set contribution to disease. This is expected with microbes that produce of toxins or inflammatory molecules that work in a dose dependent manner. Because of this, the kPath formula described above would not be accurate for Salmonella, however, it could be modified to simply be a metric like the PPT value without the consideration of mortality: Eq 4
Using literature values [22–27] we calculated the Pathogenic Potential (PP), Pathogenic Potential in regards to time (PPT), and Fs/T for (A-C) Listeria spp., (D-F) Salmonella enterica, (G-I) Staphylococcus aureus, (J-L) Pseudomonas aeruginosa, and (M-O) Group B Streptococcus. Overall, we found various relationships between the measures of pathogenic potential and the bacterial inoculum that varied species to species. While most of the (A, D, J) PP values had a negative exponential relationship with the inoculum and are best-fitted with an exponential decay function, S. aureus had positive exponential relationships between the (G) PP and (H) PPT versus inoculum, and (M,N) Streptococcus and (K) P. aeruginosa (PPT only) had positive linear relationships between the PP and PPT versus inoculum, best-fitted with a simple linear regression. All the bacterial species investigated besides (F) Salmonella enterica had a linear Fs/T vs. I relationship, which is inconsistent with what is seen in fungi. The linear relationship indicates each bacterium influences the degree and speed of death, rather than the order of magnitude of bacteria.
There was also variation between the PP vs I and PPT vs I relationships in bacteria, where the relationships were positive and linear, as opposed to the negative exponential ones in the fungi we analyzed (Fig 6A, 6B, 6D, 6E, 6F, 6G, 6H, 6J, 6K, 6M and 6N). This would suggest that in infections of these species (S. aureus., P. aeruginosa, and Streptococcus spp.) that each additional bacterium causes a set unit of damage, whereas for fungi, there are diminishing returns with increasing inoculum with regards to damage from each additional fungal cell. There does not seem to be an association between the positive linear PP, PPT, and Fs/T relationships and whether the bacteria are Gram-negative or Gram-positive. However, this pattern would suggest there is a dose-dependent effect causing death in the G. mellonella larvae, such as the secretion or production of a toxin or inflammatory molecule (Table 1).
PP, PPT, and Fs/T of entomopathogenic nematodes in G. mellonella
G. mellonella are common models for infection with entomopathogenic nematodes, including the purpose of culturing the nematodes and even using them as bait to collect nematode species in the wild. We calculated the PP, PPT, and Fs/T for two entomopathogenic nematode species  in G. mellonella. The PP and PPT vs I relationships, like those seen in C. neoformans, L. monocytogenes, and Salmonella enterica, manifested a negative exponential trend, with some variability in the middle inoculum infections (Fig 7A, 7B, 7D and 7E). The Fs/T vs I curve was positive and roughly linear, although it has a sigmoidal shape, closely fitted by an exponential one phase decay line (Fig 7C and 7F). It is worth noting these nematodes themselves do not kill the insect larvae. Once the larvae are infected with the nematodes, the nematodes release bacteria that are highly pathogenic and encode toxins that kill the host.
Using literature values , we calculated the PP, PPT, and Fs/T for the entomopathogenic nematodes (A-C) Heterorhabditis spp. strain Hgj and (D-F) Steinernema carpocapsae strain mg1. Generally, there were exponential PP vs. I and PPT vs. I relationships (as fitted by a one phase exponential decay function), as seen with fungi and some bacteria, with some variation in the middle-inoculum groups. The (C,F) Fs/T vs I relationships were best fitted by a one phase exponential decay (exponential plateau) function.
PP of the G. mellonella Nuclear Polyhedrosis Virus (GmNPV)
We calculated the PP of the G. mellonella Nuclear Polyhedrosis Virus (GmNPV), which is a baculovirus that primarily infects Lepidoptera. The results of Stairs’ 1965 study , yielded a clear negative exponential relationship between PP and inoculum of virus, whereas the data from Fraser and Stairs’ 1982 study  yielded an inverted U-shaped curve with an exponential negative relationship at the higher viral inoculum (Fig 8).
The pathogenic potential of the GmNPV (nuclear polyhedrosis virus) was calculated from published values [29,30] and plotted against inoculum. There is a negative exponential relationship between the amount of virus used to infect G. mellonella and the pathogenic potential in the Stairs 1965 study. In the Fraser and Stairs 1982 study, the relationship is varied, where the lower inocula have a positive exponential relationship with pathogenic potential, and the higher inocula have a negative exponential relationship with PP. Both plots are fitted with an exponential one phase decay function.
Modeling relationships between pathogenicity and inoculum
After noting various relationships between the pathogenicity metrics (PP, PPT, Fs/T) and inoculum we sought to understand how these differences occurred. Hence, we modeled PP, PPT, and Fs/T calculations for a hypothetical microbe at different inoculum (Fig 9). For one microbe, we calculated the Fs value as a direct function of the inoculum, represented by Eq 5 and 6, where x1 and y1 represent variables dependent on the mortality, Fs, T, and I of the infection (Eq 5.1 and 6.1). For the purposes of Fig 9, we used x = 10−5 and y = 105.Eq 5Eq 5.1Eq 6Eq 6.1
Modeled PP (A), PPT (B), and Fs/T (C) values using linear-based methods of calculating Fs and T (black data points) or log-based methods of calculating Fs and T (pink data points), or a mix of both (green data points), as described by the formulas in the graph key. Example organisms that fall under each category are listed below their respective group. PP and PPT values are fitted with a one phase exponential decay function. The linear based Fs/T values (black and green points) are fitted using a simple linear regression, whereas the log-based values (pink points) are fitted using a semi-log line.
Plotting the PP, PPT, and Fs/T values revealed a pattern similar as expected (Fig 9, black data points). For the second microbe, we aimed to model disease progression based on the magnitude of the inoculum, and in doing so, used Eq 7 and 8, where x2 and y2 represent variables dependent on the mortality, Fs, T, and log(I) (Eqs 7.1 and 8.1). For the purposes of Fig 9, we used x = 0.1 and y = 10.Eq 7Eq 7.1Eq 8Eq 8.1
This resulted in PP, PPT, and Fs/T values that when plotted yielded negative exponential PP and PPT and a positive logarithmic Fs/T, such as C. neoformans and B. bassiana (Fig 9, pink data points).
Calculating PP, PPT, and Fs/T across microbes for the same infectious inoculum
Through the fitted exponential or linear lines for the PP, PPT, and Fs/T versus I plots, we are able to use the equations of the line to calculate theoretical PP, PPT, and Fs/T values for infectious inoculums that have not yet been experimentally studied. This provided a way to compare measures of pathogenicity amongst microbes, even when the original experiments are performed at different inoculum. These calculated values are found in Table 2. It is worth noting that these values are preliminary, and based on literature, and should not be taken as definitive until experimentally confirmed using the exact inoculum. It can, however, be used to approximate disease severity outcomes when planning experimental design.
The concept of Pathogenic Potential (PP) was spawned from the notion that all microbes have some capacity to cause disease if acquired by a host in sufficient numbers. Disease occurs when the host has incurred sufficient damage to affect homeostasis and host damage can come from direct microbial action (e.g., toxins), the host immune response, or both (Casadevall & Pirofski 1999). According to this view, no microbes can be unambiguously labelled as either pathogens or non-pathogens, since pathogenicity is dependent on inoculum, host immunity, and other factors that affect the outcome of the host-microbe interaction . In this work, we experimentally derived values for the PP and PPT for the fungi Cryptococcus neoformans in the invertebrate model organism Galleria mellonella and analyzed literature data with our mathematical equations. This analysis revealed deep differences between pathogenic microbes that are interpreted as reflecting different type of virulence mechanisms. To place this work in the context of discovery, we rely on the process of “seeking new laws” proposed by Richard Feynman for how of the laws of nature are identified . Previous papers have imagined the concept of Pathogenic Potential [1,2], or as Feynman would say, these works have “guess[ed] it,” which he describes as the first step in seeking new laws to describe the natural world . In this work, we undertook the next step, which according to Feynman, is to, “compute the consequences of the guess,” or in other words, to experimentally determine the guess’ validity, then further expand the comparisons to additional “real-world” experiential observations. Following the insight of Feynman on the discovery of natural laws, this work can be considered the next step whereby the experimental work is done to to confirm or disprove the yet-to-be established theoretical equations. The current data supports the insight that microbes have diverse relationships between Pathogenic Potential and inoculum.
Insights for Pathogenic Potential versus fungal inoculum
For C. neoformans, we investigated how the PP and PPT correlated with the infective inoculum moth larvae. We found that infections with smaller inocula had a larger PP and PPT, despite fewer host deaths (Fs and M values) and longer survival times (T). Further, this relationship was exponential, meaning that the PP and PPT values increased exponentially with decreasing inoculum. While this result may seem counterintuitive because lower inoculum would be expected to produce less severe disease in infected larvae, it makes sense when considering the survival data. For example, almost 40% of the larvae infected with 103 cells of C. neoformans died, while less than twice as many (~75%) died from the larvae infected with ten times as many cells (104). Thus, the average fungal cell in a lower inoculum infection contributes more towards death than fungal cells in a higher inoculum infection. This relationship may be exponential because in many microbes, proliferation and growth are exponential, as evident by the doubling of yeast cells during reproduction. Although immune defenses could reduce the growth rate in vivo, microbial survivors would still grow exponentially albeit at lower replication rates. If the pathogenicity of a microbe is related to microbial burden within tissues, then it makes sense that the relationship between signs and symptoms, mortality, and pathogenicity and the initial inoculation concentration are also exponential relationships rather than simple linear ones.
For the purposes of this work, we calculated the Fs value using mortality of the larvae due to the consistency of mortality being reported in literature reports and the fact that mortality is an easily measured outcome of infection. We note that G. mellonella can exhibit other signs and symptoms of infection, including systemic melanization and reduction in movement, which could be used to calculate a Fs value independent of morality. These values are occasionally reported, but more widespread reporting of signs and symptoms of infection could be helpful in providing more nuanced calculations in the future. In applying these concepts, Fs should be defined by what is most appropriate for the microbe and host depending on the measurable outcomes of the specific host-microbe interaction.
The relationship between Fs/T and inoculum, For C. neoformans infections the experimental data for the relationship between Fs/T and inoculum was logarithmic. Unlike the relationship between PPT and inoculum, the Fs/T value increased with increasing inoculum but plateaued as inoculum increased. This makes intuitive sense since the value of Fs/T roughly equates to the number of individuals with signs of disease or deaths over time. Plotting the linear relationships of Fs vs. inoculum and Fs/T vs. inoculum allowed us to derive the minimum inoculum required to cause disease and death. These relationships for C. neoformans infection in G. mellonella larvae were generally conserved in mammalian models of infection using different mouse backgrounds and through different inoculation routes. Our calculated LPI for C. neoformans was one order of magnitude lower for intravenous infection than intranasal infection, which may be reflective of the extra physical and immunological barrier of the respiratory mucosa. The consistency of results between mice and moths suggests that C. neoformans causes disease in a similar manner in both hosts, and that the resulting relationships are due to a property of the fungus and/or the immune response, suggesting a conserved mechanism of virulence. In mammals the inflammatory response to C. neoformans can contribute to host damage , while in moths, infection can trigger widespread melanization, which could also damage tissues .
The PP and PPT analysis revealed the importance of comparing results from experiments performed using the same inoculum, especially when comparing the difference in pathogenicity of different strains of the same microbial species, or when comparing a mutant strain to the wildtype. Comparing different PP and PPT derived from experiments using different inoculum could cause the ΔPP to be off by orders of magnitude depending on the nature of the curve. However, we also demonstrate how pathogenicity data collected using different inocula can be compared by fitting Fs/T versus I plots thus providing new options for comparative analysis. Our results provide support for the view the capacity for virulence is relative, such that labelling a microbe a pathogen under one set of circumstances does not mean the microbe is equally as pathogenic under a separate set of circumstances. PP and PPT themselves are not intrinsic and immoveable statements on the absolute pathogenicity of a microbe, but rather provide a way to holistically and situationally evaluate pathogenicity given specific factors and variables. The PP would also change in the setting of an infection treated with an effective antimicrobial agent, where the expected Fs and M values decrease, T increases, and I (at the beginning of therapy) remains constant, and as such, changes in PP and PP T following treatment could be used to measure therapeutic efficacy. Conversely, immunosuppressive treatments or conditions that broadly enhance host susceptibility to infection would lead to increased PP and PP T, which can then be used to identify infection-related the risks involved in certain treatments.
We used published data of G. mellonella infection with other microbes to analyze PP vs. I and PPT vs. I relationships, and found that the linearity of the relationship varied, depending on the microbe. Fungi such as B. bassiana, nematode species, GmNPV virus, and some bacteria manifested an exponential negative relationship between PP and I, while some other bacteria, namely Streptococcus and Staphylococcus, had linear positive relationships between PP and I, indicating that each bacteria contributes directly to pathogenicity in a fixed and measurable amount. Similar trends are seen when we evaluated the PPT vs. I relationship.
Development of the Pathogenic Constant kPath
The slope of the linear relationship Fs/T and log(I) was defined as kPath. The kPath provides a new way describe the relationship between all the components of pathogenic potential (morbidity, time until onset of mortality, and inoculum) in a manner that is constant at any inoculum and can thus allow for comparisons of pathogenicity between different strains or isolates where the experiments were performed at different inoculum–a comparison that cannot be fairly made using other pathogenic potential metrics. A high kPath would indicate a highly pathogenic microbe, as each additional microbe results in a steep increase in disease and death over time, while a low kPath would indicate a relatively weak microbial pathogen. Additionally, for microbes that cause disease through growth and persistence within tissues, a high kPath could be associated with fast microbial doubling times, whereas low or zero kPath could be associated with microbes that have slower rates of growth within the host. A kPath of zero could also indicate that the microbe is not pathogenic or that the outcome is not dependent on the initial infective inoculum. When this is not the case, as it may not be with H. capsulatum or P. lutzii, it could indicate that the starting inoculum is irrelevant to disease either because of the presence of a potent toxin that is equally effective in low doses as it is in high doses, or an irregular and slow growth within the host. We note that for some for H. capsulatum the values the kPath had a negative sign, which would indicate less severe disease from increasing inocula. While we caution on drawing conclusions from this experimental data until confirmed, it is possible that in some infectious diseases that a threshold inoculum is needed to trigger effective immunity to control infection, which could result in negative kPath values. Additionally, data extracted from mouse literature indicate there is a positive Fs/T vs. I relationship and a positive kPath value for P. brasiliensis and H. capsulatum. This underscores the importance of comparing data within the same host, and the possibility that the same microbe could have different mechanisms of causing to disease in different hosts. This can then in turn affects the relationship between PP, PPT, and Fs/T versus I relationships. In some microbes, predominantly in bacteria, the relationship between Fs/T and I is linear and not logarithmic. For these microbes, the kPath would be defined differently, and instead rely on the direct inoculum itself. The linear kPath equation could be used to compare bacterial virulence in similar ways between different strains and inoculums. Interestingly, the kPath of C. neoformans in G. mellonella was nearly the same as it was in mice, again, consistent with the notion that C. neoformans behaves similarly in murine and Gallerian host immune systems with regards to virulence. The lines of best fit for PP vs. I, PPT vs. I, and Fs/T vs. I could be used as a method to roughly predict disease progression and pathogenicity of certain infectious inoculums. This could be helpful for planning experimental design, where a certain disease progression or pathogenicity may be desired for the conditions tested (i.e., antimicrobial drug efficacy during a mild infection).
Insights into pathogenesis from PP, PPT and Fs/T versus I relationships
The relationships between parameters of pathogenicity developed here (PP, PPT, Fs/T, kPath) provide new potential insights into how the organism cause disease and death within the host. If the microbe has a positive linear relationship in the PP vs I, PPT vs I, or Fs/T vs I plots, it is consistent with the notion that disease and death primarily result from increasing microbial burden, such that each additional microbial cell causes a proportional increase in host damage that when cumulative would result in the death of the host. This could be a pathogen that damages the host directly through the production of toxic substances or indirectly by eliciting a tissue-damaging inflammatory response that kills the host in a dose-dependent manner or that the host mounts a tissue-damaging inflammatory response that is dependent on microbial burden or a combination of both. The two microbes with the most consistent linear positive relationship were Staphylococcus aureus and Streptococcus spp., both of which are known to produce a large suite of toxins during infection [34,35]. Conversely, for a microbe that has a negative exponential PP vs I or PPT vs I relationship, with a positive logarithmic Fs/T vs I, the magnitude of starting inoculum makes a large contribution to the outcome of the host-microbe interaction and the severity of any ensuing disease. For these microbes, growth and survival in the host determines disease severity, and abundant growth within the host causes death. Microbes that fall under this category included C. neoformans, which produces virulence factors such as melanin, polysaccharide capsule, and urease that predominantly allow the fungus to persist and survive within the host rather than intoxicate the host. Consistent with this view, cryptococcosis tends to be a chronic disease that kills the human host after months of slow and progressive damage in the brain, often mediated by increased intracranial pressure resulting from fungal proliferation .
In contrast, microbes that produce virulence factors that help survival within the host and damage the host tissues directly (C. albicans with candidalysin, adhesins, and proteases), have mixed patterns in their PP, PPT, and Fs/T vs I relationships. C. albicans has no clear PP or PPT vs I relationship, which may be indicative of complex pathogenesis, where it produces a smattering of virulence factors that induce host damage, such as serine aspartyl proteases, candidalysin, and confronts the host with both hyphal and yeast cells [37–42], biofilms, and multiple adhesins [41,43–46]. For C. albicans, the mixture of the damage and persistence-type virulence factors could cause no clear PP vs I relationship. C. albicans does not have a clear correlation between PP/PPT and inoculum but does have a positive logarithmic Fs/T vs I relationship, suggests that a mix of host damage and host survival factors may play a role in determining PP and PPT, but the positive logarithmic Fs/T values are determined more by the replication and growth of the fungus within the host.
Overall, we note remarkable heterogeneity in the relationships between PP, PPT, I, and Fs/T for various microbes with one host, Galleria mellonella. We also note that the similarities observed for C. neoformans curves with G. mellonella and mice suggests commonalities between the interaction of this fungus with a mammalian and insect host, respectively, and hint that certain patterns may be conserved. We consider this study a preliminary exploration of a complex topic, but we note that it is discriminating amongst pathogenic microbes and provides new insights into the problem of virulence. We caution that the results described here involved mostly involved data in the G. mellonella host, which lacks an adaptive immune response. Furthermore, we caution that insights gathered from analysis of literature data came from different research groups, which carries the potential for considerable inter-laboratory experimental variation. While we find similarities between PP, PPT, and Fs/T versus I C. neoformans infections in murine and Gallerian hosts, a more detailed understanding of the commonalities and differences in host-microbe interactions will require detailed studies in other systems. This is especially the case with human infections, where there is tremendous variability in immune systems, underlying conditions, and environmental variables within the global population that would require nuanced studies and analysis.
In summary, we use the pathogenic potential equations to identify new and unexpected relationships between important variables in the study of microbial pathogenesis such as Fs, I and T. The differences observed here in PP vs. I and Fs/T, imply differences in pathogenesis that are likely to reflect different strategies to survive within the host, promote their own dissemination, and cause host damage over time. For example, if a microbe causes damage through growth and survival, the order of magnitude (log) inoculum would likely be the relevant determining factor of disease (i.e. logarithmic Fs/T vs. I relationship. Whereas if the microbe causes damage through toxins or lytic proteins, pathogenicity would likely be directly dependent upon each microbial cell (i.e. positive linear Fs/T vs. I relationship) (Summarized in Fig 10). Explaining the differences in the shapes and signs of the PP vs. I, PPT vs. I, and Fs/T curves suggests new avenues for research that could provide fresh insights into the problem of virulence.
The top panel indicates microbes, such as Streptococcus and Staphylococcus that produce toxins that have a dose-dependent effect on survival. This results in positive PP vs. I and a positive linear Fs/T vs I relationships. In the bottom panel is Cryptococcus which has an exponential negative relationship with PP vs. I and a logarithmic positive relationship with Fs/T vs. I, which we propose is because Cryptococcus causes host death through fungal burden, which would be a log-based relationship between starting inoculum and disease, rather than a dose-dependent linear one.
Materials and methods
G. mellonella last-instar larvae were obtained from Vanderhorst Wholesale, St. Marys, Ohio, USA. Cryptococcus neoformans strain H99 (serotype A) and Candida albicans strain 90028 were kept frozen in 20% glycerol stocks and subcultured into Sabouraud dextrose broth for 48 h at 30°C prior to each experiment. The yeast cells were washed twice with PBS, counted using a hemocytometer (Corning, New York, USA), and adjusted to the correct cell count.
Infections of Galleria mellonella
Last-instar larvae were sorted by size and medium larvae, approximately 175–225 mg, were selected for infection. Larvae were injected with 10 μl of fungal inoculum or PBS control. Survival of larvae and pupae was measured daily through observing movement with a physical stimulus.
Literature survey for calculating PP, PPT, and Fs/T for other microbes
We performed a literature search using combinations of the search terms “Galleria mellonella,” “inoculum,” “Kaplan-Meier,” “LT50,” “10^4, 10^5, 10^6,” along with the specific name of the microbe or murine strain we were interested in investigating further. PP, PPT, and Fs/T were calculated from literature that used G. mellonella as a model to study various infectious diseases using the following criteria: (1) the survival of at three inoculums were measured for each microbe, (2) the survival data was measured with enough time resolution to see the individual Kaplan-Meier survival curve (3) there was clear data that had overall mortality of the larvae (i.e. an appropriate y-axis to estimate percent mortality), and (4) there was at least a reported LT50 (median survival time) or a Kaplan-Meier curve (with the exception of the GmNPV data) in order to calculate the T and Fs values. Overall, we analyzed data from sixteen papers which mostly fit our criteria. There are other examples in literature that could be used, however, many do not test more than three inoculums, have host survival data with insufficient temporal resolution to accurately determine median survival, or do not report median survival time. For the purposes of this paper, Fs value were calculated as the total mortality of larvae or mice since cumulative incidence for the signs or symptoms of infections in G. mellonella and mice are often unreported or inconsistent in literature.
Statistical analysis and regressions
Linear and non-linear regressions were performed using GraphPad Prism Version 8.4.3. Simple linear regressions were used for the linear regressions. Both semi-log non-linear regressions and one-phase exponential decay non-linear regressions were used. Regression method used is described in the figure legend. For some graphs, the 95% confidence interval was plotted, as calculated by the GraphPad Prism software. Equations of the line used for theoretical PP, PPT, and Fs/T values were generated by GraphPad and calculated using Microsoft Excel.
S1 Fig. Pathogenic Potential of Paracoccidioides brasiliensis and Histoplasma capsulatum in Murine Hosts.
Using literature values [19–21], we calculated the pathogenic potential (PP), pathogenic potential in respect to time (PPT), Fs/T, lowest pathogenic inoculum (LPI), and kPath for P. brasiliensis (A-D) and H. capsulatum (E-H) in mice. Lines in (D, H) indicate linear regressions. LPI means calculated lowest pathogenic inoculum.
S1 Table. Spreadsheet with all the Fs, inoculum, T, and M data used for calculating the PP, PPT, and Fs/T for this work.
We would like to thank Ella Jacobs and Dr. Quigly Dragotakes for their feedback during the writing process for this manuscript. Fig 10 was made using BioRender.com.
- 1. Casadevall A. The Pathogenic Potential of a Microbe. mSphere. 2017;2: e00015–17. pmid:28251180
- 2. Casadevall A. Expanding the Pathogenic Potential Concept To Incorporate Fulminancy, Time, and Virulence Factors. mSphere. 2022 [cited 26 Jan 2022]. pmid:35080468
- 3. Casadevall A, Pirofski LA. Host-pathogen interactions: redefining the basic concepts of virulence and pathogenicity. Infect Immun. 1999;67: 3703–3713. pmid:10417127
- 4. Smith D F. Q., Casadevall A. Fungal immunity and pathogenesis in mammals versus the invertebrate model organism Galleria mellonella. Pathogens and Disease. 2021;79. pmid:33544836
- 5. Hussein KA, Abdel-Rahman MAA, Abdel-Mallek AY, El-Maraghy SS, Joo JH. Pathogenicity of Beauveria bassiana and Metarhizium anisopliae against Galleria mellonella. Phytoparasitica. 2012;40: 117–126.
- 6. Vertyporokh L, Hułas-Stasiak M, Wojda I. Host–pathogen interaction after infection of Galleria mellonella with the filamentous fungus Beauveria bassiana. Insect Science. 2020;27: 1079–1089. pmid:31245909
- 7. Vertyporokh L, Wojda I. Immune response of Galleria mellonella after injection with non-lethal and lethal dosages of Candida albicans. Journal of Invertebrate Pathology. 2020;170: 107327. pmid:31945326
- 8. Li D-D, Deng L, Hu G-H, Zhao L-X, Hu D-D, Jiang Y-Y, et al. Using Galleria mellonella–Candida albicans Infection Model to Evaluate Antifungal Agents. Biological and Pharmaceutical Bulletin. 2013;36: 1482–1487. pmid:23995660
- 9. Thomaz L, García-Rodas R, Guimarães AJ, Taborda CP, Zaragoza O, Nosanchuk JD. Galleria mellonella as a model host to study Paracoccidioides lutzii and Histoplasma capsulatum. Virulence. 2013;4: 139–146. pmid:23302787
- 10. Scorzoni L, de Paula e Silva ACA, Singulani J de L, Leite FS, de Oliveira HC, Moraes da Silva RA, et al. Comparison of virulence between Paracoccidioides brasiliensis and Paracoccidioides lutzii using Galleria mellonella as a host model. Virulence. 2015;6: 766–776. pmid:26552324
- 11. Medoff G, Maresca B, Lambowitz AM, Kobayashi G, Painter A, Sacco M, et al. Correlation between pathogenicity and temperature sensitivity in different strains of Histoplasma capsulatum. J Clin Invest. 1986;78: 1638–1647. pmid:3782474
- 12. Marcos CM, Tamer G, de Oliveira HC, Assato PA, Scorzoni L, Santos CT, et al. Down-regulation of TUFM impairs host cell interaction and virulence by Paracoccidioides brasiliensis. Sci Rep. 2019;9: 17206. pmid:31748561
- 13. Gomes JAR. Functional and genetic analysis of sexual reproduction in the thermodimorphic fungal pathogen Paracoccidioides spp. 2017 [cited 14 Feb 2022]. Available: http://repositorium.sdum.uminho.pt/
- 14. Yamaguchi M, Ohkusu M, Biswas SK, Kawamoto S. Cytological Study of Cell Cycle of the Pathogenic Yeast Cryptococcus neoformans. 日本医真菌学会雑誌. 2007;48: 147–152. pmid:17975529
- 15. Ali MF, Tansie SM, Shahan JR, Seipelt-Thiemann RL, McClelland EE. Serial Passage of Cryptococcus neoformans in Galleria mellonella Results in Increased Capsule and Intracellular Replication in Hemocytes, but Not Increased Resistance to Hydrogen Peroxide. Pathogens. 2020;9: 732. pmid:32899539
- 16. Kashino SS, Calich VLG, Burger E, Singer-Vermes LM. In vivo and in vitro characteristics of six Paracoccidioides brasiliensis strains. Mycopathologia. 1985;92: 173–178. pmid:4088290
- 17. Wojda I, Jakubowicz T. Humoral immune response upon mild heat-shock conditions in Galleria mellonella larvae. Journal of Insect Physiology. 2007;53: 1134–1144. pmid:17631308
- 18. Wojda I, Kowalski P, Jakubowicz T. Humoral immune response of Galleria mellonella larvae after infection by Beauveria bassiana under optimal and heat-shock conditions. Journal of Insect Physiology. 2009;55: 525–531. pmid:19232408
- 19. Pedroso VSP, Vilela M de C, Santos PC, Cisalpino PS, Arantes RME, Rachid MA, et al. Development of a murine model of neuroparacoccidioidomycosis. Journal of Neuroparasitology. 2010;1. Available: https://www.cabdirect.org/cabdirect/abstract/20113192125
- 20. Adamson DM, Cozad GC. Effect of Antilymphocyte Serum on Animals Experimentally Infected with Histoplasma capsulatum or Cryptococcus neoformans. Journal of Bacteriology. 1969;100: 1271–1276. pmid:5361216
- 21. Sá-Nunes A, Medeiros AI, Nicolete R, Frantz FG, Panunto-Castelo A, Silva CL, et al. Efficacy of cell-free antigens in evaluating cell immunity and inducing protection in a murine model of histoplasmosis. Microbes and Infection. 2005;7: 584–592. pmid:15820152
- 22. Six A, Krajangwong S, Crumlish M, Zadoks RN, Walker D. Galleria mellonella as an infection model for the multi-host pathogen Streptococcus agalactiae reflects hypervirulence of strains associated with human invasive disease. Virulence. 2019;10: 600–609. pmid:31230520
- 23. Martinez MR, Wiedmann M, Ferguson M, Datta AR. Assessment of Listeria monocytogenes virulence in the Galleria mellonella insect larvae model. PLOS ONE. 2017;12: e0184557. pmid:28898264
- 24. Mukherjee K, Altincicek B, Hain T, Domann E, Vilcinskas A, Chakraborty T. Galleria mellonella as a Model System for Studying Listeria Pathogenesis. Applied and Environmental Microbiology. 2010 [cited 16 Feb 2022]. pmid:19897755
- 25. Bender JK, Wille T, Blank K, Lange A, Gerlach RG. LPS Structure and PhoQ Activity Are Important for Salmonella Typhimurium Virulence in the Gallleria mellonella Infection Model. PLOS ONE. 2013;8: e73287. pmid:23951347
- 26. Sheehan G, Dixon A, Kavanagh K 2019. Utilization of Galleria mellonella larvae to characterize the development of Staphylococcus aureus infection. Microbiology. 165: 863–875. pmid:31107207
- 27. Hill L, Veli N, Coote PJ. Evaluation of Galleria mellonella larvae for measuring the efficacy and pharmacokinetics of antibiotic therapies against Pseudomonas aeruginosa infection. International Journal of Antimicrobial Agents. 2014;43: 254–261. pmid:24361354
- 28. Kumar P, Ganguly S, Somvanshi VS. Identification of virulent entomopathogenic nematode isolates from a countrywide survey in India. International Journal of Pest Management. 2015;61: 135–143.
- 29. Stairs GR. Dosage-mortality response of Galleria mellonella (Linnaeus) to a nuclear-polyhedrosis virus. Journal of Invertebrate Pathology. 1965;7: 5–9.
- 30. Fraser MJ, Stairs GR. Susceptibility of Trichoplusia ni, Heliothis zea (Noctuidae), and Manduca sexta (Sphingidae) to a nuclear polyhedrosis virus from Galleria mellonella (Pyralidae). Journal of Invertebrate Pathology. 1982;40: 255–259.
Feynman R. The Character of Physical Law. Cambridge, MA, USA: MIT Press; 1967.
- 32. Pirofski L-A, Casadevall A. Immune-Mediated Damage Completes the Parabola: Cryptococcus neoformans Pathogenesis Can Reflect the Outcome of a Weak or Strong Immune Response. mBio. 2017;8: e02063–17. pmid:29233901
- 33. Kan H, Kim C-H, Kwon H-M, Park J-W, Roh K-B, Lee H, et al. Molecular Control of Phenoloxidase-induced Melanin Synthesis in an Insect. J Biol Chem. 2008;283: 25316–25323. pmid:18628205
- 34. Barnett TC, Cole JN, Rivera-Hernandez T, Henningham A, Paton JC, Nizet V, et al. Streptococcal toxins: role in pathogenesis and disease. Cellular Microbiology. 2015;17: 1721–1741. pmid:26433203
- 35. Otto M. Staphylococcus aureus toxins. Current Opinion in Microbiology. 2014;17: 32–37. pmid:24581690
Chayakulkeeree M, Perfect JR. Cryptococcosis. In: Hospenthal DR, Rinaldi MG, editors. Diagnosis and Treatment of Human Mycoses. Totowa, NJ: Humana Press; 2008. pp. 255–276. https://doi.org/10.1007/978-1-59745-325-7_14
- 37. Moyes DL, Wilson D, Richardson JP, Mogavero S, Tang SX, Wernecke J, et al. Candidalysin is a fungal peptide toxin critical for mucosal infection. Nature. 2016;532: 64–68. pmid:27027296
- 38. Schaller M, Korting HC, Schäfer W, Bastert J, Chen W, Hube B. Secreted aspartic proteinase (Sap) activity contributes to tissue damage in a model of human oral candidosis. Molecular Microbiology. 1999;34: 169–180. pmid:10540295
- 39. Hube B, Sanglard D, Odds FC, Hess D, Monod M, Schäfer W, et al. Disruption of each of the secreted aspartyl proteinase genes SAP1, SAP2, and SAP3 of Candida albicans attenuates virulence. Infection and Immunity. 1997;65: 3529–3538. pmid:9284116
- 40. Sanglard D, Hube B, Monod M, Odds FC, Gow NA. A triple deletion of the secreted aspartyl proteinase genes SAP4, SAP5, and SAP6 of Candida albicans causes attenuated virulence. Infection and Immunity. 1997;65: 3539–3546. pmid:9284117
- 41. Wächtler B, Wilson D, Haedicke K, Dalle F, Hube B. From Attachment to Damage: Defined Genes of Candida albicans Mediate Adhesion, Invasion and Damage during Interaction with Oral Epithelial Cells. PLOS ONE. 2011;6: e17046. pmid:21407800
- 42. Wilson D, Naglik JR, Hube B. The Missing Link between Candida albicans Hyphal Morphogenesis and Host Cell Damage. PLOS Pathogens. 2016;12: e1005867. pmid:27764260
- 43. Fukazawa Y, Kagaya K. Molecular bases of adhesion of Candida albicans. Journal of Medical and Veterinary Mycology. 1997;35: 87–99. pmid:9147268
- 44. Silva-Dias A, Miranda IM, Branco J, Monteiro-Soares M, Pina-Vaz C, Rodrigues AG. Adhesion, biofilm formation, cell surface hydrophobicity, and antifungal planktonic susceptibility: relationship among Candida spp. Front Microbiol. 2015;6. pmid:25814989
- 45. Hasan F, Xess I, Wang X, Jain N, Fries BC. Biofilm formation in clinical Candida isolates and its association with virulence. Microbes and Infection. 2009;11: 753–761. pmid:19409507
- 46. Bain JM, Lewis LE, Okai B, Quinn J, Gow NAR, Erwig L-P. Non-lytic expulsion/exocytosis of Candida albicans from macrophages. Fungal Genetics and Biology. 2012;49: 677–678. pmid:22326419
- 47. Hagihara M, Housman ST, Nicolau DP, Kuti JL. In Vitro Pharmacodynamics of Polymyxin B and Tigecycline Alone and in Combination against Carbapenem-Resistant Acinetobacter baumannii. Antimicrobial Agents and Chemotherapy. 2013 [cited 2 Mar 2022]. pmid:24277022
- 48. Anguissola S, McCormack WJ, Morrin MA, Higgins WJ, Fox DM, Worrall DM. Pigment Epithelium-Derived Factor (PEDF) Interacts with Transportin SR2, and Active Nuclear Import Is Facilitated by a Novel Nuclear Localization Motif. PLOS ONE. 2011;6: e26234. pmid:22028839
- 49. Kaufman-Francis K, Djordjevic JT, Juillard P-G, Lev S, Desmarini D, Grau GER, et al. The Early Innate Immune Response to, and Phagocyte-Dependent Entry of, Cryptococcus neoformans Map to the Perivascular Space of Cortical Post-Capillary Venules in Neurocryptococcosis. The American Journal of Pathology. 2018;188: 1653–1665. pmid:29929915