Resistance to experimental infections with Haemonchus contortus in Romanov sheep

Institut National de la Recherche Agronomique; 1 Station de Virologie et d’Immunologie, Centre de Recherches de Jouy-en-Josas, 78350 Jouy-en-Josas; 2Station d’Amédioration Génétique des Animaux, Centre de Recherches de Toulouse, BP 27, 3i326 Castanet-Tolosan Cedex; sLaboratoire des Groupes Sanguins, Centre de Recherches de Jouy-en-Josas, 78350 Jouy-en-Josas; 4Laboratoire de Génétique Biochimique, Centre de Recherches de Jouy-en-Josas, 78350 Jouy-en-Josas, France


INTRODUCTION
Since the publications of Warwick et al (1949), Whitlock (1955Whitlock ( , 1958 and Whitlock and Madsen (1958), the existence of a genetic variability in the resistance to Haemonchus contortus has been shown in several studies: the heritability estimates range around 0.25-0.30 (Le Jambre, 1978;Albers et al, 1984Piper, 1987).
As there are almost no genetic correlations between the resistance and various production traits (Alberts et al, 1984(Alberts et al, , 1987Piper, 1987), selection on resistance to H contortus would be possible and economically justified in conditions where this type of parasitism leads to large productivity losses (Holmes, 1986). However, it does not seem to be possible to use the response to an experimental infection as a largescale selection criterion because of the difficulties of such an experimentation. It would therefore be interesting to identify resistance predictors, either immunological traits or genetic markers (Courtney, 1986;Alberts and Gray, 1987;Cabaret and Gruner, 1988).
From a genetic point of view, the main objective of the present experiments was to confirm or invalidate this hypothesis. Because the typing of animals in the major histocompatibility system ( OLA) was performed retrospectively, a search for relationships between resistance to H contortus and the OLA marker was also included in this study.
From a parasitological point of view, the experimental goals were to supply additional information on the following phenomena: repeatability of faecal egg counts between successive infections, relationship between egg counts and self-cure, relationship between egg counts and degree of anaemia, acquisition of immunity to the parasite by previous contact with the parasite and effect of anthelmintic treatment on this acquired immunity.
The experiment was designed so as to give responses to questions in the fields of genetics and parasitology.

Animals
Several studies have shown that females develop stronger immunity against H contortus than males (Colglazier et al, 1968;Yazwinski et al, 1980;Luffau et al, 1981a;Courtney et al, 1985;Watson, 1986): hence only females were used in the present study, ie 51 female lambs of the Romanov breed born from 8 sires and 36 dams. The breeding animals were chosen according to their haemoglobin genotype. All sires were Xb AB heterozygotes. The dams belonged to genotypes Hb AA, Hb AB or Hb BB. The 51 lambs fell into 3 groups of 17, each representing 1 of the 3 haemoglobin genotypes. The number of animals in the 3 haemoglobin genotypes was balanced within each sire progeny so as to reduce risks of confusion between a possible haemoglobin genotype effect and a possible sire effect.
Fifty lambs and 24 of their 35 dams were typed for antigens of the OLA system. The sires were not typed but their genotypes could be inferred and transmission of markers determined in many cases.
The experimental female lambs were chosen so as to form a group as homogeneous as possible for age, weight, maintenance conditions and health in order to reduce uncontrolled factors of variation. The animals were maintained on a grass free diet from birth to avoid environmental exposure to H contortus.
Typing methods for haemoglobin and OLA systems Haemoglobin types were determined by electrophoresis (Nguyen and Bunch, 1980). Class I antigens of the major histocompatibility system were tested by the microcytotoxicity method on blood lymphocytes; the test was carried out over a period of 2h 30 min (Cullen et al, 1985). Lymphocytes of each animal were tested with 120 antisera against 22 provisional specificities, &dquo;OLA-P&dquo;. Nine haplotypes, each carrying 1 or 2 specificities, were identified in the tested animals.
Immunization experiments with aggregated human serum albumin The 51 experimental lambs were immunized at the age of about 6 months with heat aggregated human serum albumin (HSA: 200 mg/animal) by intravenous injection.
Their serum was collected before and 14 d after the administration of the antigen, titred by passive haemagglutination using red blood cells tanned and sensitized with HSA (Weir, 1978). The technique used to determine the serum agglutination titre has been described previously (Nguyen, 1984).

Experimental infections with H contortus
According to various studies, sheep develop immunity against H contortus from the age of about 7 months (Jarrett et al, 1961;Manton et al, 1962;Urquhart et al, 1966a, b;Knight and Rodgers, 1974;Wilson and Samson, 1974;Benitez-Usher et al, 1977;Duncan et al, 1978;Riffkin and Dobson, 1979;Smith and Angus, 1980). Our experiments therefore began when the lambs were about 8 months old. During the experimental infections, lambs were kept in well controlled conditions: open sheepfold fitted with a slatted floor, diets based on compound feed concentrates, hay and straw ad libitum. Five infection experiments were conducted successively using 3-week old larvae. Animals were infected with larvae obtained by faecal cultures according to the method of FJS Robert and PJ O'Sullivan and collected with Baerman's apparatus (Luffau et al, 1981a, b). The required number of larvae were counted microscopically and suspended in 20 ml of ordinary water. This suspension was administered orally. The strain maintained at the Station of Virology and Immunology was supplied initially by Professors GM Urquhart and EW Allonby (Glasgow).

Experiment 1
In experiment 1, lambs were divided into 3 groups: -18 animals were given 3 infections successively: a primary infection on DO with 5 000 larvae, a secondary one on D32 with 10 000 larvae and a 3rd one on D64 with 20 000 larvae (group 1); -18 animals were given 2 infections successively: a primary infection on D32 with 10 000 larvae and a secondary one on D64 with 20 000 larvae (group 2); -15 animals were given an infection of 20 000 larvae on D64 (group 3).
The 3 groups were formed so as to obtain a balanced distribution of the various paternal origins and haemoglobin genotypes. Each measure (number of eggs per gram) was the mean egg count of 3 different samples. These egg counts were good indicators of the worm burdens of the animals (Roberts and Swan, 1981).
The number of red blood cells (per pl of blood) was determined by measuring the variation in the potential difference (Celloscope 401 -Ljungberg -Stockholm, Sweden) induced by the passage of red blood cells (blood dilution 1/800) in an electric field. The apparatus was periodically checked according to the microscopical method of Malassez.
For measuring the haemoglobin content (g/100 ml blood), haemoglobin of the red blood cells lysed by saponin was fixed and transformed into cyanmethaemoglobin.
The haemoglobin content was measured by spectrophotometry (absorption at 630 nm).

Experiment 2
The surviving 49 animals were divided into 2 groups, irrespective of the group they were part of in experiment 1: -the 26 animals of group 1 were not drenched prior to experiment 2; hence they were carriers of a residual H contortus population; -before starting experiment 2 the 23 animals of group 2 were drenched with a highly effective anthelmintic, Thibenzole MSD powder (thiazolyl benzimidazolethiabendazole ND, Paris, France).
In these 2 groups, each animal was given 10 000 larvae on DO of experiment 2 (263 days after DO of experiment 1). Faecal egg counts were made on the following 20 dates: Dl, D0, D17, D20, D24, D27, D31, D34, D38, D41, D45, D48, D56, D59, D80, D83, D88, D91, D95 and D98.  The immunological, parasitological and haematological variables used are given in table I. The parasitological variables were defined from decimal logarithms of mean egg counts over certain periods (in order to normalize distributions and obtain more homogeneous variances). The choice of periods was based on the kinetics of faecal eggs counts in the successive infection experiments.
Thus, in each of the 3 groups of experiment 1, a peak faecal egg count was observed after the primary infection (fig 1). This peak was located from D24-D37 in group 1, from D56-D57 in group 2 and from D88-D107 in group 3: the PRIMPEAK variable reflects this peak.
In groups 1 and 2 of experiment 1, the secondary infection was followed by a very large drop in faecal egg counts (from D39 to D46 in group 1 and from D74 to D81 in group 2): this was the classical self-cure phenomenon. Variable SELFCURE reflects this phenomenon; it is defined as the difference between the primary peak and the depression subsequently to the self-cure. A secondary peak could be observed immediately after this depression (D46 to D53 in group 1 and D84 to D88 in group 2): variable SECPEAK reflects this peak. In experiments 2, 3, 4 and 5, the faecal egg counts increased after the infection ( fig  2). Variables PEAKEXP2, PEAKEXP3, PEAKEXP4 and PEAKEXPS reflect the high egg counts after the infection (from D27-D48 in experiment 2, D26-D54 in experiment 3, D26-D55 in experiment 4 and D28-D52 in experiment 5). The synthetic variable PEAK2,35 is the mean of the 3 variables previously defined in expriment 2 and in its 2 replications, ie experiments 3 and 5 also involving 2 groups of animals (a group drenched before infection and a non-drenched group). The synthetic variable PEAK235 does not include experiment 4 in which all animals were drenched prior to infection.
The haematological parameters are defined as means of given measures over certain periods. The choice of periods here is again based on a kinetic examination. The number of red blood cells, the packed cell volume and haemoglobin content decreased during the period corresponding to the primary egg count peak: from D23-D39 in group 1, D53-D67 in group 2 and D88-D102 in group 3 (figs 3a, b, c). Variables RBCPRIM, PCVPRIM and HCPRIM, respectively account for this decrease in the 3 previously cited parameters.

Factors
The factors of variation considered are given in table II. Two of these factors (HBALLELE, the haemoglobin allele received from the sire and OLALLELE, the OLA haplotype received from the sire) are nested within sire. According to analyses, the response to immunization with HSA was considered as a variable or a factor.

Method of analysis
Analysis of the humoral immune response Two methods were used for the statistical analysis of the humoral immune response, ie x 2 test and analysis of variance.
Chi-square tests of independence were carried out between the RESPOND factor (accounting for the immunization &dquo;responder&dquo; or &dquo;non-responder&dquo; character) and various other factors of variation of table II (sire, haemoglobin genotype and OLA , haplotypes).
Analysis of variance were performed on variable ANTIHSA, accounting for the immune response to aggregated human serum albumin (table III). The number of experimental animals was not large enough to make an analysis simultaneously including all factors of variation; there would have been numerous either empty or low cells. Accordingly, several analysis of variance models were used each involving a small number of factors. This can also be applied to the analyses of variance of the parasitological and haematological variables, treated in the following paragraphs.
Analysis of the parasitological and haematological variables of experiment 1 Table IV shows the analysis of variance models applied to the parasitological and haematological variables of experiment 1. When the factors did not include RE-SPOND (immunization &dquo;responder&dquo; or &dquo;non-responder&dquo; trait) or TITRE (category of anti-HSA antibody titre), the ANTIHSA variable (reflecting the humoral response) was added in order to study its correlation with the parasitological and haematological variables. The same procedure was used for analysis of the variables of experiments 2, 3, 4 and 5.
Analysis of the pana,sitological variables of experirrzents 2, 3, 4 and 5 Table V gives the models of the analyses of variance performed on the parasitological variables of experiments 2, 3, 4 and 5. Analyses of variables of experiments 2, 3 and 5 included necessarily factor GROUP235 corresponding to the group (a group drenched before infection, a non-drenched group). This was not the case for analyses of the variable of experiment 4, since in this experiment all animals were given the anthelmintic treatment before infection.

Analysis of various parasitological variables considered as repeated measures of the same character
A new approach consists of considering that the parasitological variables PRIM-PEAK, PEAKEXP2, PEAKEXP3, PEAKEXP4 and PEAKEXP5 (referring to experiments 1, 2,. 3, 4 and 5, respectively) constitute repeated measures of the same parasitological overall variable OVERALL. Table VI gives models of analyses of variance on the OVERALL variable. Each model includes necessarily: -the EXPGROUP factor corresponding to the experiment and group combination in which the OVERALL variable was measured; -the ANIMAL factor corresponding to the experimental animal in which the measure was made.

Analysis of the humoral immune response
Chi-square tests of independence between the &dquo;responder&dquo; character and various other factors (sire, haemoglobin genotype and OLA haplotypes) No x 2 was significant at the 0.05 level, with the exception of the test of independence between the &dquo;responder&dquo; character and the OLA -P14+9 haplotype (xi! = 5.953, significant at the 0.02 level) : this x 2 resulted from a preferential association between the &dquo;responder&dquo; character and the OLA -P14 + 9 haplotype.
Analyses of variance of the ANTIHSA variable reflecting the humoral immune response Among the 23 analyses of variance described in table III, only analyses no 8 and 19 showed a significant effect at the 0.05 level; in both cases, it was the effect of the OLA -P14 + 9 haplotype whose presence in the studied sample was related to an increase in the level of anti-HSA antibodies.
Analysis of the parasitological and haematological variables of experiment 1 Table VII summarizes the results of the analyses of variance whose models are described in table IV. Table VIII gives the residual correlations calculated on the parasitological and haematological variables relative on the primary peak of experiment 1 and on the immunological ANTIHSA variable. Table IX gives the residual correlations calculated on all parasitological variables of experiment 1 (relative to the primary peak, secondary peak and to the self-cure) as well as on the immunological ANTIHSA variable.
Analysis of the parasitological variables of experiments 2, 3, 4 and 5  Figure 4 illustrates the effect of the TITRE factor (anti-HSA antibody titre) on the overall parasitological OVERALL variable.

DISCUSSION
Phenotypic relationships between the various parasitological variables A highly significant (P < 0.001) positive residual correlation was observed in experiment 1 between the variables reflecting the primary and secondary peak faecal egg counts. In contrast, these 2 variables were negatively correlated with the variable reflecting the self-cure, the correlation being only highly significant (P < 0.001) between the self-cure and the secondary peak: in other words, the animals with the lowest faecal egg counts were also those that best expelled their parasites.
Positive (generally significant) residual correlations were observed in the successive experiments 2, 3, 4 and 5 (table XI) between the variables accounting for the egg counts. The highly significant effect of the ANIMAL factor on the overall parasitological OVERALL variable (table XII) illustrates the repeatability of the mean egg output during the peaks of the 5 successive infection experiments.
Phenotypic relationships between faecal egg counts and degree of anaemia There were highly significant (P < 0.001) correlations between the mean number of red blood cells per mm 3 of blood, the average packed cell volume and the mean level of haemoglobin during the primary peak of eggs passed in experiment 1 (table VIII). The PRIMPEAKvariable, reflecting the peak faecal egg counts during primary infection was negatively correlated with the 3 haematological variables. This corresponds to the phenomenon of anaemia classically associated to large faecal egg counts (Whitlock, 1955(Whitlock, , 1958Evan et al, 1963;Pradhan and Johnstone, 1972;Altaif and Dargie, 1978a, b;Roberts and Swan, 1982;Albers et al, 1984).

Effect of primary dose of infective larvae of faecal egg counts and anaemia
The factor GROUPl (group in experiment 1) had a significant effect on all parasitological and haematological variables measured during the primary peak of experiment 1 (table VII): as expected, the larger the larval intake, the higher the faecal egg counts and the degree of anaemia (figures 1 and 3).
Effect of vaccination on immunity to the parasite Considering the kinetics of faecal egg output in group 1 of experiment 1 (figure 1): the peak faecal egg counts after the 1st infection (with 5000 larvae) was higher than the peak after the 2nd infection (with 10 000 larvae) which was higher than the peak after the 3rd infection (with 20 000 larvae). Likewise, in group 2 of experiment 1 (figure 1), the primary peak exceeded the secondary peak although the 2nd dose of infective larvae was 2-fold higher than the 1st one (20 000 larvae instead of 10 000).
With the same dose of infective larvae (10 000 larvae on D32 or 20 000 larvae on D64), animals which had previously experienced infection with H contortus reacted by eliminating fewer eggs than those infected with the parasite for the first time.
These observations (based on figure 1 and confirmed statistically by analyses of variances not shown here) illustrate the phenomenon of immunity to the parasite (ie protection) acquired by &dquo;vaccination&dquo;, ie by previous infection with the parasite (Clunies Ross, 1932;Luffau, 1975;Luffau et al, 1981a, b).
Effect of anthelmintic treatment on immunity to the parasite In experiments 2 and 3, the group drenched before infection eliminated significantly more eggs than the non-drenched group (figure 2 and first line of table X); the anthelmintic treatment substantially reduced the immunity acquired previously by contact with the parasite. The phenomenon was more marked in experiment 3 which was a replication of experiment 2. In experiment 5, the same trend was observed (figure 2), but the difference between the 2 groups was not significant (table X).
Thus, total elimination of residual worms by anthelmintic treatment prior to DO of the previous experiment (experiment 4) seemed to have reduced the difference between the 2 groups.
All these results, which show the reduction of immunity to the parasite after anthelmintic treatment, confirm those obtained by Benitez-Usher et al (1977) according to whom application of such a treatment after vaccination with irradiated larvae lowered the immunity to the parasite.
Phenotypic relationships between resistance to parasitism and humoral immunity . Factors relating to the anti-HSA antibody titre (RESPOND and TITRE) had significant effects on various parasitological variables of experiments 1 (table VII), 2, 3, 4 and 5 (table X) as well as the overall parasitological OVERALL variable (table XII). The higher the production of anti-HSA antibodies the larger the faecal egg counts, as shown by adjusted means in figure 4. This positive relation between faecal egg count and humoral immunity is also illustrated by the positive coefficients of correlation between various parasitological variables reflecting the faecal egg output and the variable ANTIHSA reflecting the humoral immunity (tables IX and XI).
Contrary to the hypothesis put forward by Cuperlovic et al (1978), response to an immunization with aggregated human serum albumin is not a predictor of resistance to H contortus. This negative correlation between resistance to helminths and response to an immunization has already been observed in mice (Blum and Cioli, 1978;Deelder et al, 1978;Perrudet-Badoux et al, 1978;Wakelin, 1978). In sheep, Albers et al (1984) did not find any significant correlation between resistance to H contort and response to an immunization with chicken red blood cells.
Sire effect on resistance to H contortus Significant sire effects were evidenced on the PRIMPEAK variable accounting for the primary peak faecal egg counts in experiment 1 (table VII) and on the overall parasitological OVERALL variable pertaining to all experiments (table XII): these effects were significant at the 0.05 and 0.10 level, respectively. The number of experimental animals (51 offspring of 8 sires) was too small to make a heritability estimation. However, our results are in favour of a sire effect on resistance; they are in keeping with those of Le Jambre (1978), Albers et al (1984 and Piper (198?) who found a heritability ranging from 0.25-0.30 for resistance to H contortus. Relationships between haemoglobin system and immunological, parasitological and haematological variables Neither the HBGENO factor (haemoglobin genotype) or the HBALLELE factor (haemoglobin allele received from the sire) had any significant effect at the 0.05 level on the immunological or parasitological variables, although the experiment was designed to verify the existence of such effects. These findings do not agree with those of other authors (Evans et al, 1963;Jilek and Bradley, 1969;Radhakrishnan et al, 1972;Allonby and Urquhart, 1976;Dargie, 1976, 1978a, b;Preston and Allonby, 1979;Dally et al, 1980;Luffau et al, 1981a, b;Courtney et al, 1985), but they are in keeping with the results of Le Jambre (1978), R,iffkin and Dobson (1979), Courtney et al (1984), Riffkin and Yong (1984) and Albers and Burgess reported by Piper (1987), who did not find any relationship between resistance to H contortus and haemoglobin system. Thus, although our data do not lead to detection of any relationship of humoral responsiveness (to human serum albumin) or of resistance to H contortus with the haemoglobin system, there is evidence of a statistically significant effect (P < 0.05) of the HBALLELE factor (haemoglobin allele received from the sire) on the mean packed cell volume (table VII). Hence a relationship between haemoglobin system and post-infection degree of anaemia cannot be excluded. According to the adjusted means, it seems that animals carrying the HbA allele were less anaemic than the others. These results are in agreement with those of Evans and Whitlock (1964), Radhakrishnan et al (1972), Dargie (1976, 1978a, b) and Albers and Burgess reported by Piper (1987). The post-infection differences observed between animals of various haemoglobin genotypes might simply be due to differences existing in non-infected animals (Agar et al, 1972). These differences might arise from oxygen affinity differences between haemoglobins A and B. Haemoglobin A has a higher oxygen affinity: at equal pressure, it releases less oxygen, which might cause the creation of compensatory mechanisms in haemoglobin A carriers (Agar et ad, 1972).
Relationships between the OLA system with immunological, parasitological and haematological variables The results obtained show statistically significant effects of various OLA haplotypes on the humoral response as well as on the faecal egg counts and the degree of anaemia after parasite infections (tables VII, X and XII). Thus, we cannot exclude the existence, within or close to the OLA system, of genes affecting these various phenomena. These results disagree with those of Cooper et al reported by Piper (1987), who did not find any association between OLA system and resistance to H contortus. But they do agree with those of Outteridge et al (1984, 1985, 1986, 1987 and 1988) who found an association between the OLA system and the response to a vaccination against Trichostrongylus colubrifo!rmas. A relationship between the major histocompatibility complex and resistance to nematode parasites has also been demonstrated in the case of the guinea pig-Trichostrongylus colnbriformis system (Geczy and Rothwell, 1981) and the mouse-Trichostrongylus spiralis system (Wassom et al, 1979).

CONCLUSION
In terms of parasitology, the results obtained lead to a more accurate determination of a certain number of phenomena such as repeatability of faecal egg counts between infections, negative relationship between faecal egg count peaks and self-cure intensity, positive relationship between faecal egg counts and degree of anaemia, acquisition of immunity by previous contact with the parasite and reduction of this immunity by anthelmintic treatment.
In terms of genetics, the results invalidate the hypothesis that homozygous sheep carriers of haemoglobin A have lower faecal egg counts than the others as well as the hypothesis that animals with the best humoral immune response are the most resistant to parasitism. On the other hand, they do not exclude the hypothesis that genes within or close to the OLA system might affect the resistance to H contortus. The latter conclusion, in keeping with those of other studies, emphasizes the role of the OLA system as a potential marker of resistance to parasitism.