Microbiological, virological, bacteriological, immunological, medical, epidemiological, historical, anecdotal

Tag: Journal of Experimental Medicine

Data update: How dyestuffs make stuff die


This year, antibiotic researcher Mark Wainwright published a “Discussion” in the International Journal of Antimicrobial Agents, suggesting that we should take another look at a subject that he works on but most scientists lost interest in 6 decades ago: highly toxic synthetic molecules that can be used as local antiseptics and have distinctive bright coloration. What he calls “photoantimicrobials” represent a subset (those that can be targeted more precisely, because they are activated by light) of therapeutic molecules that earlier generations simply called flavines, anilins, or “dyes”.

Wainwright tells us:

Browning introduced the application of the basic dyes named above [acriflavine, crystal violet and brilliant green] to battlefield wounds in 1914/15. The aim was always to achieve antisepsis and healthy wound closure, and the approach was also used in pre-operative sterilisation. The dyes used thus represent local antiseptics. Importantly, however, they were an improvement on the available hypochlorite solutions owing to the fact that they were not inactivated by fluids associated with wounds, such as serum. Fleming’s rebuttal of this work, noted above, was based purely on laboratory measurement. However, the disinfection of real wounds and the concomitant recovery of patients could hardly be argued against.

By far the majority of clinical work involving dyes was carried out before 1945. At this juncture, the penicillins and other antibiotics or natural product-derived antibacterial agents became the driving force in infectious diseases therapy. From the distance of the early 21st century, it is apparent that these wonder drugs have not been used correctly and that their overuse has allowed a much more rapid development of resistance mechanisms than might otherwise have been the case.

In terms of modern healthcare, dye therapy might be considered obsolete and ineffective, and this would indeed be the case if the performance criteria were still the same as in Browning’s day. However, recent research has developed a different approach, one that uses locally applied dyes in conjunction with light to provide a microbial killing effect. Such dyes may be referred to as photodyes or, more properly, photosensitisers. Given their remarkable utility in the field under discussion, they are commonly called photoantimicrobials.

The photoantimicrobial effect has been known in the laboratory since the turn of the last century, but has only been investigated realistically since the early 1990s and is currently proposed foruse in oral and ENT (ear, nose and throat) disinfection. Its remarkable activity lies in the in situ production of reactive oxygen species (ROS) on illumination. The highly reactive nature of species such as singlet oxygen, the hydroxyl radical and the superoxide anion ensures rapid oxidative damage to simple cells, sufficient to guarantee cell death. More importantly, there are no known microbial resistance mechanisms able to combat ROS produced in this way.

Furthermore, among the lead compounds used in photo antimicrobial discovery are the same dyes used by Ehrlich and Browning; methylene blue, crystal violet and acriflavine.

Brilliantly colored solutions for treating localized infections and washing out wounds may be making a comeback. And not just to treat pets, for which acriflavine, for example, is now used. In preparation for that retro craze, let’s look at some of the original pre-Depression research. churchman-gazette

Most of it is in German, which I will ignore.

Much of the non-German work on “dye therapy” was led by urologic surgeon John W. Churchman of Yale University. At the top of this post is a rare color photo of bacteria in a Petri dish from the time when Julius Richard Petri (1852-1921) was still alive. This comes from a 1913 Journal of Experimental Medicine paper by Churchman, The Selective Bactericidal Actions of Stains Closely Allied to Gentian Violet, whose figures are otherwise black & white. Even the limitless moneybags of Old Elihu and the Rockefeller Institute only extended to printing one color figure.

* * *

Churchman’s papers are not very quantitative, but other “dye therapy” researchers did large-scale comparisons of multiple colored antibiotics, at multiple concentrations, against multiple bacterial species.


On January 1, 1914, the Journal of Experimental Medicine published the Observations of Josephine S. Pratt and Charles Krumwiede Jr. (not to be confused with their Further Observations, published 4 months later). They established a method for testing large numbers of drug/bacteria combinations while minimizing the amount of agar needed, by putting two samples in each Petri dish, slanted away from each other. Here they describe the process in vague terms that would be better presented as a 5-minute video.

As a routine a batch of agar was selected which was found especially suitable for the more feebly growing [bacteria]. To the hot agar an appropriate volume of a watery solution of dye was added to give the final dilution desired. The same agar was used in each experiment.

The most convenient and economical method was found to be as follows. One unopened Petri dish was used to tilt up one side of a second dish. In the opposite side was poured just sufficient agar to give a satisfactory slant. This dish was covered and used to tilt up a third dish, and so on in a row. After the agar had set, slants were poured in the other side of the Petri dishes which were tilted in the reverse direction. In this way two mixtures of agar could be used in the same dish, very little agar being required for each slant.

After inoculating all these plates with bacteria, they used the eye test. Compared to a control plate with no dyestuffs, how much did the dye prevent the bacteria from growing after 18 to 24 hours? Not at all? Was it “restrained”? “Markedly restrained”? Completely eliminated?

Here’s Table 1.


That is a lot of information. 11 drugs, tested against 30 bacteria (12 Gram-positive and 18 Gram-negative), makes 330 combinations, of which only 7 potential tests were not performed. (I think the ellipsis means “not performed”.) Each of these 323 tests was done at 3 concentrations, making 969 data points in a single table.

To turn this into graphs would require many graphs. It would also require us to turn the X’s and +’s into a semiquantitative system (let’s say + = 4, ± = 3, X = 2, —* = 1, and — = 0). And since each data point is restricted to those 5 possibilities, you wouldn’t gain much by looking at a bunch of individual dots or bars anyway.

Let’s leave it as a table.

The main flaw of this table is that all the symbols look similar, except “—” . The bottom half of the table is a wall of symbols for different degrees of growth. It’s hard to see trends, because the symbols are hard to distinguish from each other, so it looks like every bacterium is resistant to every drug. Which is not quite true.

Instead of these text symbols, how about something that may be more intuitive — representing the 5 levels of growth as 5 colors. This may not be intuitive in every culture, and it may not work for the color-blind or in B&W printouts, but I took the liberty of changing the table. Now as bacteria proliferate, they pass from “no growth” (white) through yellow, green, and purple stages until they attain “growth like control” (charcoal grey). At first I wanted to range from white to black, but if the cells are black you couldn’t see the lines between them. Or maybe you can’t see them anyway.

For the tests that weren’t done, I replaced the ellipses with neutral grey. And finally, the two types of “diphtheroid bacilli” produced exactly the same results, so they were conflated into one.


So where does this data lead us? Krumwiede and Pratt draw limited conclusions. The tables speak for themselves. To briefly summarize their Summary section: the “streptococcus-pneumococcus group” is more dye-resistant, and the “dysentery bacillus group” is highly unpredictable with dye-resistance showing “no correlation with the common differential characteristics”. Also, they discovered mutations that make bacteria lose resistance. (“Among Gram-negative bacteria a strain is occasionally encountered which will not grow on violet agar, differentiating it from other members of the same species or variety.”) Finally, in the middle of the Summary section they say this.

The reaction is quantitative, although the quantitative character is more marked with some species than with others.

Now, a modern reviewer would look unkindly on that sort of admission. First, it seems like it’s not quantitative, it’s semiquantitative. Instead of measuring the number of colonies, you’re grading “growth” on a scale from 1 to 5. And which are the species that don’t have a “quantitative character”? Why don’t you use some other method to see how well those ones are growing? And by the way, how repeatable is your eye test? And what does it mean exactly? Let’s see examples of “growth like control”, “restrained growth”, and “markedly restrained growth”. Does it mean there were fewer colonies, or the colonies were smaller, or both? And why use the same 3 concentrations of all 11 drugs? Maybe 1:500,000,000 would have still killed the bacteria, and be less toxic to patients.

That’s what I’d say, if the journal hadn’t told me “Your review is now 100 years late and consequently is no longer needed.”

* * *

One more paper: from 7 years later, Gay and Morrison in the January 1921 Journal of Infectious Diseases. Whereas Krumwiede & Pratt (1914) was the first in their series, this is a sequel.


Krumwiede & Pratt took a limited range of dyes, and used then on every sort of bacteria they could find. Now Gay & Morrison use every dyestuff they can find on a limited range of bacteria. This is a much longer paper, so I’ll just address their first set of data, about bacterial growth in culture, and ignore their innovative rabbit empyema model.

Table 1:


In this table, the numbers mean the reciprocal of the minimum inhibitory concentration (MIC). The MIC is the smallest concentration of the drug that will prevent bacteria from growing. If the MIC is 5 ng/ml, the bacteria should grow if the concentration is lower than that, and the bacteria should die if the concentration is 5 mg/ml or greater.

“2,000” means a 1 to 20,000 dilution. I think that’s weight/volume (1 gram in 20,000 milliliters). The lower the number in this graph, the more concentrated the drug needs to be to kill bacteria. A drug marked “2,000” needs to be a thousand times more concentrated to have the same effect as a drug marked “2,000,000”.

This table is clear. Intuitively we look at the numbers and recognize that 2,000 is smaller than 20,000 and therefore represents less of a dilution. If they were in scientific notation, it wouldn’t be as intuitive.

The only problem is the use of “0”. The most concentrated dyes they used were at a 1:2,000 concentration. If the bacteria still grew, they listed this drug as “0”, or not active against the bacteria. Taken literally, “0” means Staphylococcus would grow on media consisting entirely of methylene green. I would just write the word “inactive” instead of the number 0.

* * *

I might also expand the table to include the rest of the 40 dye molecules Gay & Morrison used in this study. That’s right — in addition to the 12 molecules listed above, they have data on Acid fuchsin, Acid violet, Acridine orange, Azo-acid red, Basic fuchsin, Benzo-azurin, Brilliant cresyl blue, Columbia blue R, Congo red, Crystal ponceau, Cyanin B, Diamil blue, Erioglaucin A, Janus dark blue B, Methylene blue (medical purity), Methylene blue GG, Neutral red, New fast green 3B, Nile blue, Oxamin violet, Rhodulin violet, Safranin, Sauer grün, Scarlet 6R, Setocyanin, Sulphon acid blue R, Toluidin blue, and Wasser blau.

But most of the data is in paragraph form. And it’s not immediately clear.


Let’s make an expanded version of the table above, containing all the dyestuffs. Even the dyestuffs that never killed any of the bacteria. And make some other changes:

  • Keep them arranged in order of effectiveness, but put the most effective ones at the top.
  • In fact, divide them into categories. Those that inhibited all 3 bacterial species, those that inhibited 2, those that inhibited 1, and those that were completely ineffective.
  • Update the nomenclature. We consider Bacillus typhosus to be Salmonella now.
  • And… why not color-code the chart. Maybe blue dyes will do one thing, and red dyes will do something else. Might as well include that information. Patterns may emerge.


Isn’t it nice to see those colors? And it shows how much easier it is to kill Streptococcus pyogenes. There’s only one dye that kills either Staphylococcus or Salmonella but fails to kill S. pyogenes. And it shows that blue and red dyes aren’t very antibacterial, while green ones kill bugs dead. Why is that?

Data update: Destroyed by freezing does not equal alive

In the experiments reported in the present paper a number of active agents, some undoubtedly living, others equally unquestionably not living, and still others of a doubtful nature, were subjected to repeated freezing (-185°C.) and thawing. By these tests it has been possible to determine that mere destruction or inactivation of a substance cannot be accepted as proof that it existed in a living state.

In 1926, legendary virologist / bacteriologist Thomas M. Rivers addressed the still-extant question “Are viruses alive?” through the simple experimental method of freezing them over and over. The paper in question came out in the Rockefeller Institute house publication Journal of Experimental Medicine (vol.45[1]: pp. 11-21). Not exactly a gripping title.


What kind of data is in here? Well, for most of the graphs they take an “active agent” (either bacteria, virus, or enzyme), freeze-thaw it “as often as desired”, and then see if it’s still active. The data is pretty straightforward, with a separate section for colon bacilli, a section for Virus III, a section for “a bacteriophage lytic in colon bacilli”, and so on. But it’s never summarized in a way that compares the different “agents” to each other. Let’s try to do that.

The low temperature (-185°C.) used in the experiments to be reported was obtained by means of commercial liquid air which was transported from the plant to the laboratory in Dewar flasks. Desired amounts of the air were transferred to deep Dewar beakers where small amounts of the substances to be frozen, enclosed in Noguchi tubes, were completely immersed for several minutes. After the substances had been completely frozen they were quickly thawed in tap water (16-18°C.).

What does “active” mean for these various substances? How did they determine that freeze-thawing had destroyed the substance’s activity? This was different for each substance.


Bacteria can be measured the same way we measure them now, by making serial dilutions and plating each dilution on agar, then seeing how many colonies grow. Bacteriophage can be measured the same way, by first making a “lawn” (agar plate fully covered with bacteria) and applying different dilutions of bacteriophage, then seeing how many “plaques” (holes in the lawn) are formed by the phage killing the bacteria.

Complement can be measured by seeing how long it takes to destroy “given amounts of red blood cells in the presence of a great deal of amboceptor“.

Trypsin can be measured somehow, Dr. Rivers doesn’t specify except by saying that his colleague Dr. Northrop took care of that part of the study. Nowadays Dr. Northrop would be a co-author. But the paper would then have to list Rivers as both the first author and the last author somehow, since Northrop didn’t do enough to merit either status.

Anyway, to find “details of the technic” we are directed to a paper by Northrop and Hussey (1923), showing a very clever method by which a solution of gelatin is exposed to trypsin, and at different timepoints the gelatin’s viscosity is measured.


And how do you measure viscosity? With a viscosimeter.


I’m having trouble understanding sentences like “The gelatin-water time ratio was approximately 3”, but the point is that you can measure the amount of trypsin by measuring how fast it turns gelatin into runny gelatin. Nowadays you would use a colorimetric assay, in which trypsin cuts the protein that has some sort of colored label attached to it, and you would measure how much colored label gets released into solution.

* * *

Finally, the three mammal viruses.

“Vaccine virus” and “Virus III” are both introduced to the skin of rabbits, probably by scarifying and then rubbing the virus into the scratches. They look for a “virus reaction” in the skin surrounding where the virus was inoculated, and measure this semi-quantitatively based on how bad of a sore forms. I hope they aren’t simply measuring the immune response to the inoculation, because even killed virus should produce some immune response.

“Vaccine virus” is basically what we now call vaccinia virus. “Virus III” is a more interesting question.

All stocks of Virus III were lost some time before the invention of electron microscopy. Nobody can now be sure what exactly this rabbit-specific virus was, but it was probably Leporid herpesvirus 2, as described by Nesburn in 1969. For an objective summary of the Virus III story, read the page on LHV-2 (p. 380) in The Laboratory Rabbit, Guinea Pig, Hamster, and Other Rodents (Academic Press, 2012).

“Herpes virus” (Herpes simplex) activity is measured differently. They inject HSV into rabbit brains, and they look for dead rabbits. It seems like this assay should actually be quantitative, based on looking at how long it takes the rabbits to die, but all their rabbits either died within a week or lived longer than a month, so the results were clear without measuring time to death.

 * * *

The results are pretty simple, which is why I would like a summary figure instead of a series of tiny individual tables for each thing they studied.


Some comments on the data:

  • “Locke’s solution” is like what we call Ringer’s solution, the intravenous fluid given to people who have suffered blood loss. This page indicates that it’s the same as Ringer’s solution, but buffered by bicarbonate instead of lactate. But most other sources indicate that it also includes glucose, making it closer to Tyrode’s solution.
  • Locke’s solution is basically a physiological salt solution, so what’s the difference between that and the “physiological salt solution” used in the phage experiments? The latter doesn’t contain glucose, and isn’t buffered; instead of a pH of roughly 7.6, the pH is “between 6 and 7”.
  • Vaccinia virus was the most stable, retaining the ability to create a skin lesion even after 34 freeze-thaws. Virus III was destroyed after 12, and the other herpesvirus was destroyed after 24.
  • Both bacteria and bacteriophage had more than 99% of their activity destroyed by freeze-thawing in Locke’s solution. All three big viruses were more hardy than that.
  • I don’t know what a “1:10 dilution of trypsin” is. How many USP Trypsin Units is that? What’s the molarity? We have to take Dr. Northrop’s word for it, that it’s the same system for trypsin dilution he always uses.
  • At a mere 1:10 dilution, complement and trypsin were not damaged by 12 freeze-thaws. But when diluted further, they were very susceptible to freeze-thaw.
  • In fact, bacteria and bacteriophage were also more susceptible to freeze-thaw at low dilutions. Even vaccinia virus lost all its effect after 34 dilutions at 1:100,000 dilution, although it didn’t have much activity at that dilution to begin with.
  • Of further interest: the difference between high and low dilutions was only observed in Locke’s solution or salt solution, not broth. Remember these cutting-edge findings when you make your own aliquots.

Gallery: Early immunofluorescence

How did immunofluorescence begin? Who was the first scientist to use fluorescently labeled antibodies to stain cells under a microscope?

The answer to this is pretty clear. Albert Coons of Harvard Medical School, in two papers from 1942 and 1950. In this 1962 speech (published in JAMA) Wesley Spink of the University of Minnesota describes Coons’s accomplishments.

The first time fluorescent molecules were attached to antibodies was in the early 1930s, as described in a brief letter to Nature by John Marrack (1934). But at this point the antibodies are being used in suspension, for serological experiments similar to those that detect bacteria by agglutination. In this case the bacteria can be measured colorimetrically by mixing them with fluorescently labeled antibody, washing off nonspecific antibodies, and seeing if the bacteria turn pink.

In fact, here’s the entire article.


In The Demonstration of Pneumococcal Antigen in Tissues by Means of Fluorescent Antibody (1942), Coons and associates showed that fluorescein isocyanate (FIC) could be attached to antibodies and used to stain thin tissue sections on slides, in the same way many histological stains had been used in the past. Rabbits were immunized with pneumococcus, and their serum was conjugated with FIC. This was used to visualize the bacteria in the liver of a mouse with severe pneumococcus infection. The first immunofluorescence image ever published, I believe, was this 20-minute (!) exposure.


* * *

But the fluorescent antibody technique didn’t take off until Coons and Kaplan published the sequel eight years later. I can’t find any papers from the 1940s that cite the Coons paper and actually include fluorescent images of their own.

The pivotal 1950 paper is entitled Localization of Antigen in Tissue Cells: II: Improvements in a Method for the Detection of Antigen by Means of Fluorescent Antibody. Kind of like Rambo III, the numbering of this title is odd because there was no Localization of Antigen in Tissue Cells Part I. Instead, it’s explained that “the first paper in this series was entitled ‘The Demonstration of Pneumococcal Antigen in Tissues by Means of Fluorescent Antibody'”, meaning that this is a sequel 8 years in the making.

And what a sequel! Over 2,100 citations, according to Google Scholar’s notoriously inflated algorithm. Meanwhile PubMed’s claim of 301 is a clear underestimate. It was cited a lot.

Suddenly everyone was able to use fluorescent antibodies to find out where their antigen of choice was located inside the cell, or inside the tissue, or inside the animal. Especially after 1958 when Riggs and colleagues at the University of Kansas showed how to make FITC (fluorescein isothiocyanate) conjugates, which are more stable than FIC and don’t require phosgene for their chemical synthesis. Using thiophosgene is no picnic either, but apparently it’s less deadly.

* * *

Just as early issues of the Journal of Virology are filled with papers that show “The Ultrastructure of [Name of virus]” by taking random pictures of it with an electron microscope, it seems like it was easy to get a microscopy paper published in the early 1960s. You injected rabbits with a substance, labeled their serum with fluorescein, stained some slides, and put together a manuscript called “Demonstration of [Name of substance] in [Name of tissue] by the Fluorescent Antibody Technique”.

I decided to look through some papers that cite Coons and Kaplan (1950), to find examples of early immunofluorescence. Here are a dozen images, all from at least fifty years ago.

* * *


(2) Chicken muscle fiber, stained with rabbit globulin specific for myosin (0.5 microns, 1000x). (3) A “dark medium phase contrast” image of the same slide for comparison. (Finck et al., J Biophys Biochem Cytol 1956)


(3) Cottontail rabbit papilloma, stained with rabbit serum specific for Shope papilloma virus and goat anti-rabbit (75x). (4) An H&E stain of the same slide, to show that the virus antigens are in the keratinized region. (Mellors, Cancer Res 1960)


Embryonic chicken fibroblast infected for 10 hours with the Rostock strain of fowl plague virus (now known to be influenza A virus H7N1), stained with rabbit serum specific for “g antigen” (now called NP) (950x). The same cell stained by Giemsa-Wright stain, to show the nucleus. (Breitenfeld and Schäfer, Virology 1957)


Rat eye (lens and iris), stained with rabbit globulin specific for rat glomerulus (190x). Another section of the same eye stained with non-specific rabbit globulin. This is an example of the papers that used specific antiserum to find common antigens in seemingly unrelated tissues, in this case kidney and eye. (Roberts, Br J Ophthalmol 1957)


Human tissues, stained with human serum specific for blood group A or B. Yes, human serum. A volunteer of blood group A was used to get the serum against blood group B, and vice versa. (Szulman, J Exp Med 1960)


Group B streptococci, stained with rabbit globulin specific for group B streptococci (magnification unspecified). This paper’s total lack of negative controls is charmingly naive. (Moody et al., J Bacteriol 1958)


(5) Amoebae frozen during pinocytosis, with free fluorescent antibody (non-specific rabbit globulin) visible in pinocytosis vacuoles (2000x). (6) Phase-contrast view of another section of the same amoeba. (Brandt, Exp Cell Res 1958)


Mononuclear cells from nasal smears of ferrets infected with influenza (PR8 strain), stained with rabbit globulin specific for influenza virus (560x). Cells display a range of nuclear and cytoplasmic staining patterns. (Liu, J Exp Med 1955)


Cladosporium bantianum mold stained with rabbit serum specific for C. bantianum (1000x). By comparison, C. carrionii stained with serum specific for C. bantianum, to show lack of cross-reactivity. (Al-Doory and Gordon, J Bacteriol 1963)


Mixture of two yeast species photographed under both ultraviolet and visible light simultaneously (magnification unspecified). Saccharomyces cerevisiae (near center) glows green when stained by rabbit serum specific for S. cerevisiae. Red light serves as counterstain for other species (Pichia membranefaciens). (Kunz and Klaushofer, Appl Microbiol 1961)


A single myoblast isolated from a stage 23 chick embryo, stained with rabbit globulin specific for myosin (1100x). The arrow indicates where the nucleus obscures the myosin. (Holtzer et al., J Biophys Biochem Cytol 1957)


Kidney of rat injected with nephrotoxic rabbit antibodies to cause nephritis; the rabbit antibodies concentrate in the glomeruli, as seen by staining with goat globulin specific for rabbit globulin (37x and 205x). (Ortega and Mellors, J Exp Med 1956)