Complex interactions between bacteria and haemosporidia in coinfected hosts: An experiment

Abstract Hosts are typically coinfected by multiple parasite species whose interactions might be synergetic or antagonistic, producing unpredictable physiological and pathological impacts on the host. This study shows the interaction between Plasmodium spp. and Leucocytozoon spp. in birds experimentally infected or not infected with Mycoplasma gallisepticum. In 1994, the bacterium Mycoplasma gallisepticum jumped from poultry to wild birds in which it caused a major epidemic in North America. Birds infected with M. gallisepticum show conjunctivitis as well as increased levels of corticosterone. Malaria and other haemosporidia are widespread in birds, and chronic infections become apparent with the detectable presence of the parasite in peripheral blood in response to elevated levels of natural or experimental corticosterone levels. Knowing the immunosuppressive effect of corticosterone on the avian immune system, we tested the hypothesis that chronic infections of Plasmodium spp. and Leucocytozoon spp. in house finches would respond to experimental inoculation with M. gallisepticum as corticosterone levels are known to increase following inoculation. Plasmodium spp. infection intensity increased within days of M. gallisepticum inoculation as shown both by the appearance of infected erythrocytes and by the increase in the number and the intensity of positive PCR tests. Leucocytozoon spp. infection intensity increased when Plasmodium spp. infection intensity increased, but not in response to M. gallisepticum inoculation. Leucocytozoon spp. and Plasmodium spp. seemed to compete in the host as shown by a negative correlation between the changes in their PCR score when both pathogens were present in the same individual. Host responses to coinfection with multiple pathogens measured by the hematocrit and white blood cell count depended on the haemosporidian community composition. Host investment in the leukocyte response was higher in the single‐haemosporidia‐infected groups when birds were infected with M. gallisepticum. A trade‐off was observed between the immune control of the chronic infection (Plasmodium spp./Leucocytozoon spp.) and the immune response to the novel bacterial infection (M. gallisepticum).

Furthermore, responses of parasites belonging to different genera of haemosporidia to a coinfection might further depend on the pronounced differences in their life cycle as well as on their life stage.
Plasmodium infection intensity increases when corticosterone levels naturally increase in the breeding season but also when they are experimentally increased (Applegate, 1970). Thus, in house sparrows, Passer domesticus with a latent infection of Plasmodium relictum, a daily injection with corticosterone during a 10-day period in winter caused a relapse of the P. relictum infection resulting in detectable parasites in blood smears, while in control birds the parasites could not be found in erythrocytes. The natural increase in corticosterone levels at the onset of the breeding season also caused this relapse although corticosterone injections accelerated it resulting in a higher infection intensity (Applegate, 1970). Schoenle et al. (2019) similarly showed that exogenous glucocorticoids amplified Plasmodium spp. burden but less so in red-winged blackbirds Agelaius phoeniceus coinfected with Leucocytozoon spp. and/or Haemoproteus spp. By combining these results with the observation that in house finches corticosterone levels increase following an experimental infection with M. gallisepticum (Love, Foltz, Adelman, Moore, & Hawley, 2016), we can hypothesize that in birds chronically infected with Plasmodium spp., and possibly with other haemosporidia, parasitemia would increase following a M. gallisepticum infection (Dhondt & Dobson, 2017 (Sydenstricker et al., 2006), real-time polymerase chain reaction (qPCR) designed to test for the presence of the bacteria using the DNA from conjunctival swabs (Grodio, Dhondt, O'Connell, & Schat, 2008), and rapid plate agglutination (RPA) to test for the presence of M. gallisepticum-specific antibodies in blood (Sydenstricker et al., 2006).

| Testing for the presence of haemosporidia
The presence of haemosporidian parasites was tested in all birds, and the parasite lineage was identified when possible. Blood samples were taken at the same time every sampling day starting at 9 a.m. with the same bird and sampling the birds in the same sequence.
DNA was extracted from each blood sample using the DNeasy Blood & Tissue Isolation Kit (Qiagen) according to the manufacturer's instructions. To verify DNA integrity, extractions were screened on 2% agarose gels. For the molecular diagnosis, the protocol for nested polymerase chain reaction (PCR) described by Bensch, Hellgren, and Pérez-Tris (2009) was followed using the primers described in Bensch et al. (2000); Hellgren, Waldeström, and Bensch (2004); Waldenström, Bensch, Hasselquist, and Östman (2004). It targets the mitochondrial cytochrome b gene in the haemosporidian genome. The PCR was repeated three times for each sample. Given the variation among replicates in band intensity, the bands were visually scored on a scale of 0-2. All PCR products were confirmed by 2% agarose gel electrophoresis and visualized with ethidium bromide staining. Photographs of each gel were taken using a UV transilluminator with Kodak Gel Logic Digital Imaging System. All photographs were printed on the same laser printer and used to score all PCR products. Samples with the absence of a band were scored as 0; a faint band was scored as 1; and a strong band was scored as 2 (see Figure 1). For each individual, we summed the three scores to calculate a PCR score to reflect the intensity of infection. The scoring was done blind and independently by MRP and AAD. Due to our small sample size, we analyzed the PCR bands in two ways. First, we counted bands (standard method). Second, we employed the band intensity method described above. All PCR products for any haemosporidian species were sequenced and compared to the MalAvi database (Bensch et al., 2009).
Three thin blood smears were prepared per individual bird following standard techniques for haemosporidian studies (Valkiūnas, 2005) using an aliquot taken from the brachial vein.
Blood smears were air-dried, fixed with 100% methanol, and immediately stained with Giemsa stain prepared as per Petithory, Ardoin, and Ash (2005). The slides were examined using a Meiji Techno MT 4000 Biological Microscope using an oil immersion objective (100×).
Parasitemia was quantified for each smear by recording the number of infected cells in 100 random fields (Godfrey, Fedynich, & Pence, 1987) that each had approximately 200 erythrocytes for a total of 20,000 erythrocytes (red blood cells, RBCs).
The infection intensity was scored by using the number of infected RBC and by using the PCR score. The hypothesized impact of the M. gallisepticum inoculation on haemosporidian infection intensity was determined using both changes in the PCR score and changes in the number of infected RBC between day 0 (preinfection score) and day 24 postinfection (PI) and comparing birds infected with M. gallisepticum and the control group.

| Inoculation with M. gallisepticum
Roughly half of the birds were inoculated with M. gallisepticum (further: experimental birds), while the others were inoculated with Frey's medium, the standard medium in which to grow these bacteria (Kleven,2008 ) (further: controls or control birds). Given that before the inoculation, some birds were infected with Plasmodium spp., some with Leucocytozoon spp., some with neither, and some with both we assigned about half of the birds in each of the groups to each treatment. (no haemosporidia or both genera detected) were also inoculated with M. gallisepticum. In total, 18 house finches were inoculated with M. gallisepticum and 14 birds were kept as controls.

| Statistical analysis
The statistical analyses were conducted using Statistix10 (Analytical Software) and SAS v 9.3 (SAS Institute). Graphs were made using SigmaPlot 11. When distributions of residuals were not following the exponential family of probability distributions two different intervals (β 0 from 0 to 6.9, or 0 to 7.1) generated almost identical results (only the latter will be reported).

| Eye lesions in response to M. gallisepticum inoculation
Following M. gallisepticum inoculation, all house finches developed conjunctivitis in both eyes by day 3 PI. Eye scores reached a maximum by day 11 PI, and birds maintained severe signs of disease to day 24 PI (see Figure 2). In all birds in this group, M. gallisepticumspecific antibodies were detected on day 13 and day 24. None of the control birds developed lesions, and none had antibodies (Fisher's exact test: p < .0001). The mean eye score did not differ between birds infected with Plasmodium spp., Leucocytozoon spp., and both (Kruskal-Wallis ANOVA H = 5.46, df = 2; p = .07).

| Haemosporidian occurrence in the peripheral blood before and after inoculation
Before the 32 birds were inoculated with M. gallisepticum or with control medium, our triplicated PCR tests found 23 house finches to be naturally infected with a Plasmodium lineage. Seven of them were infected with PADOM11, three with WW3, and in the other 13 we did not get enough good quality DNA to be able to assign or identify the parasite lineage. Ten house finches were infected with a Leucocytozoon lineage (9 CB1, 1 CARFLA04). Two of these house finches were coinfected by Plasmodium (PADOM11) and Leucocytozoon (CB1). Out of the 32 birds, only one individual was negative for all PCR test. We did not get any sequences by PCR from infected birds that belonged to the genus Haemoproteus. In none of the birds did we detect any infected blood cells.
The number of positive PCR tests for Plasmodium spp. in replicated samples of the same individual varied between 1 (n = 4) and 3 (n = 9) and the summed PCR scores in birds that tested positive for at least one PCR test varied between 1 and 5.5 (mean 2.78 ± SE 0.247). For Leucocytozoon spp., all three tests in each bird were either positive or negative and the PCR score varied between 2.75 and 6 (4.83 ± SE 0.379).
On day 24 PI Plasmodium spp. was detected in the same individuals as before the start of the experiment, but Leucocytozoon was detected in 10 additional birds whereby two additional haplotypes (CARCHL04 and CNEORN01) were identified. Half of those additional birds were in the control group, suggesting that handling stress and captivity may have played a role in Leucocytozoon's emergence.
All 10 additional Leucocytozoon-infected birds were coinfected with Plasmodium spp., bringing the total number of birds coinfected with different haemosporidian genera to 12.
Both the band counting and the band intensity methods revealed qualitatively similar responses to M. gallisepticum inoculation (

| Quantitative responses of Plasmodium spp. and Leucocytozoon spp. in the peripheral blood to M. gallisepticum coinfection
In the experimental birds, the average number of red blood cells (RBCs) containing Plasmodium spp., that was zero before the inoculation, had increased by day 3 PI, continued to increase to day 5, reaching a maxi-

| Effects of haemosporidian coinfection
Possible effects of coinfection of parasites belonging to these two genera of haemosporidia on the response to M. gallisepticum in- that the changes in PCR scores of the two genera of haemosporidia from day zero to day 24 PI are inversely related (see Figure 4).
This result is not caused by a "regression to the mean" effect as

F I G U R E 4 Change in
Leucocytozoon PCR score plotted against the change in Plasmodium PCR score from day 0 to day 24 PI in the presence (circles) or absence (triangles) of Mycoplasma gallisepticum. All birds were coinfected with parasites belonging to two different genera of haemosporidia

| White blood cell counts
The temporal changes in white blood cell counts until day 7 PI were affected by an interaction between M. gallisepticum and haemosporidian coinfections as shown by the significant three-way interaction Period_1*MG*Haem (χ 2 = 6.88, df = 2; p = .032; Table 2

| Hematocrit
We  Note: In bold and underlined: factors left in the model. Only significant main effects on the changes with time (βs) and their lower-order interactions were left in the model. Intercepts (α 0 ) for each combination of parasites were left in the model to allow the profiles to begin as close as possible to their empirical mean (day 0). Parameter estimates are discussed in the main text, and a detailed overview is given in the Appendix.
documenting that house finch corticosterone levels naturally increased following a M. gallisepticum infection (Love et al., 2016). As hypothesized by Dhondt and Dobson (2017), we did observe a parasitemia increment in birds chronically infected with Plasmodium sp. following a M. gallisepticum infection. As Leucocytozoon spp.
is also more prevalent during the breeding season when corticosterone levels naturally increase in birds (Applegate, 1970)  contains only low copy numbers of the parasite's target sequence against a high background level of nontarget host DNA (Freed & Cann, 2006). Variation in detection might be the result of these factors. In this study, we could not determine the Plasmodium linage recovered from 13 birds. These correspond to the samples with a very low values in the PCR score after triplicate molecular diagnosis. Moreover, those birds also remained with a very low parasitemia when we examined their blood smears after the M. gallisepticum infection. It is very likely that the failure to identify the Plasmodium lineage infecting those individuals is the result of low-quality amplicons due to a very low parasitemia. Although we sequenced every amplicon we obtained from triplicate PCR, no sequences could be identified. On the other hand, every sample was tested separately for Plasmodium and Leucocytozoon. We were therefore always able to detect coinfection of these two genera.
The usefulness of repeating each PCR test three times was confirmed by differences in parasite DNA detection between replicates that are likely the result of the low parasitemia in birds in which we did not detect parasites by microscopy. The value of this approach is reinforced by the fact that following the inoculation of the birds with M. gallisepticum, it was possible to observe an increase in infection intensity for both Plasmodium and Leucocytozoon, or an increased in the probability to detect an infection (for Leucocytozoon).

| Effect of M. gallisepticum on haemosporidian load
Previous studies have shown a variety of reactions on co-occurring pathogens. Our results showed very different responses to coinfection with M. gallisepticum between the two haemosporidian genera.
Plasmodium spp. parasitemia was higher in birds coinfected with the bacterium, a result consistent with other studies in which malaria parasites and bacteria co-occur in an individual host (Rooth & Bjorkman, 1992

| Health parameters
In response to infections by a bacterium and multiple apicompl- Although the hematocrit values observed in all samples were within the normal range, and the initial decrease was probably due to the frequency of bleeding, differences in the increase in Hct values after day 7 PI were observed between treatment groups. Hct Increased more rapidly in experimental than in control birds and also differed between groups that had different coinfections. Knowing that high corticosteroids have a positive physiological effect on the erythropoiesis, we can speculate that experimental birds might have higher levels of corticosteroids due to coinfection.
From the host perspective, also diverse responses to haemosporidian infections have been observed. Thus, house martins Delichon urbicum survival was reduced when birds were infected by haemosporidian parasites, but more so when they have a double infection than a single infection (Marzal, Bensch, Reviriego, Balbontin, & De Lope, 2008); similar to this, people coinfected with malaria and dengue presented more severe clinical signs for both diseases than people infected just with one of the pathogens (Epelboin et al., 2012).
Similar to these studies we observed a positive feedback between the two pathogens in our system, coinfected hosts showed more severe clinical signs for both Plasmodium spp. parasitemia and mycoplasmal conjunctivitis. It is remarkable that some of these negative effects for the host were only observed when there was a 3-way interaction in hosts infected by two different genera of haemosporidia and the bacteriu M. Leucocytozoon spp. parasitemia only increases when Plasmodium spp. was present.
Further research will be needed to determine which immunological mechanisms cause these differences, particularly when Leucocytozoon spp. is present in the coinfection. As the physiological effects in the double-haemosporidia-infected groups were modest for both health measures, we suggest that competition among haemosporidia can ultimately benefit wild hosts under co-pathogen stress (e.g., M. gallisepticum).

| CON CLUS IONS
Our study shows complex interactions resulting from co-occurring infections in house finches. Whereby some interactions are beneficial to the pathogens (M. gallisepticum and Plasmodium spp. seem to both benefit in terms of infection intensity and thus transmission probability when co-occurring), some are antagonistic (interactions between Leucocytozoon spp. and Plasmodium spp.), and some are synergetic (effect of Plasmodium spp. on Leucocytozoon spp.).
Interactions between Plasmodium spp. and Leucocytozoon spp. in their exoerythrocytic life stages need to be studied in more detail. Birds Unlimited provided the sunflower seeds to attract and feed the birds. We gratefully acknowledge the detailed and constructive comments of two anonymous reviewers.

DATA AVA I L A B I L I T Y S TAT E M E N T
x = 0 (before manipulation); x = 1 (after manipulation); Q = day 7. α 0 : intercept, β 0 : slope of the profile curve before flexion (day 7 included), β 1 : slope of the profile after day 7.