- Email: [email protected]
A comparison of equid herpesvirus-1 (EHV-1) vascular lesions in the early versus late pregnant equine uterus
J. Comp. Path. 1996 Vol. 114, 231-247
A Comparison of Equid Herpesvirus-1 (EHV-1) Vascular Lesions in the Early versus Late Pregnant Equine Uterus K. t3. Smith, J. A. M u m f o r d and K. Lakhani Animal Health Trust, Centrefor Preventive Medicine, P O. Box 5, Newmarket, Suffolk CB8 7DW,, UK
Summary Four Welsh Mountain pony mares at 3 months of gestation and one mare at 5 months were inoculated intranasally with equid herpesvirus-1 (EHV-I: Ab4 isolate) at doses of 105 to 106.6 TCID50. All five mares became infected, but no cases of paresis or abortion occurred. On days 8, 9, 1 l, 12 (3-monthpregnant mares) and 13 (5-month-pregnant mare) after infection, a detailed examination of the pregnant uterus was made. Small numbers of vascular lesions with EHV-I antigen expression in endothelial cells were present in the uteri of the early gestational mares; thrombi were rare and loci ofthromboischaemic damage were not seen. Six pony mares previously inoculated with EHV-1 Ab4 at 9 months of gestation had a significantly greater degree of vascular abnormality than that found in the four mares infected at 3 months of gestation, but the degree of EHV-1 antigen expression and thrombosis in the uterus was similar to that found in the single mare infected when 5 months pregnant. 9 1996 W.B. Saunders Company Limited
Introduction Equid herpesvirus-1 (EHV-1) is an 0~-herpesvirus which remains a common abortigenic pathogen in Equidae despite widespread use of vaccination and management procedures designed to limit the spread of infection in susceptible populations (Allen and Bryans, 1986). Experimental studies with the highly virulent EHV-1 isolates Army 183 and Ab4, which also cause neurological disease, have demonstrated the role of endothelial cell infection and thrombosis in the endometrium of the late pregnant uterus in producing abortion (Edington et al., 1991; Smith et al., 1992, 1993). Field studies have shown that 95% of EHV-1 abortions occur in the last 4 months of gestation (Doll, 1952; Doll and Bryans, 1963). This apparent resistance of the early pregnant mare to abortigenic EHV-1 infection formed the basis of a programme of abortion control practised on studfarms in Kentucky for many years (Doll and Bryans, 1963). The programme was based on the intranasal administration of a live hamster-adapted EHV-1 isolate to broodmares in early pregnancy in order to provoke a moderate immune response with a low risk of vaccine-induced abortion. A booster inoculation was given 4 months later in order to reinforce immunity during the critical final 0021 9975/96/030231 + 17 $12.00/0
9 1996 W.B. Saunders Company Limited
232
K.C. Smith et al.
third o f p r e g n a n c y . This control p r o g r a m m e posed f o r m i d a b l e m a n a g e m e n t p r o b l e m s but r e d u c e d the incidence o f a b o r t i o n epizootics w h e n rigorously applied in successive b r e e d i n g seasons (Bryans, 1981). E H V - 1 a b o r t i o n has b e e n r e c o r d e d on rare occasions as early as the f o u r t h m o n t h o f gestation (Prickett, 1970) and, after e x p e r i m e n t a l infection, at 3 m o n t h s (Doll et al., 1955). T o investigate the low susceptibility o f the p r e g n a n t m a r e to abortigenic E H V - 1 infection in early gestation, four mares at 3 m o n t h s o f p r e g n a n c y a n d one m a r e at 5 m o n t h s were infected intranasally with the Ab4 isolate o f E H V 1, m o n i t o r e d clinically a n d virologically, a n d n e c r o p s i e d on days 8 to 13 after inoculation. T h e results were c o m p a r e d with those arising f r o m earlier experiments in which mares in late p r e g n a n c y were infected with the Ab4 isolate (Smith et al., 1992, 1993). Materials and Methods
[/grus
The Ab4 isolate of EHV-1 was obtained from a quadriplegic mare affected during an outbreak of neurological disease in 1980 (Crowhurst et al., 1981). The virus was passaged eight times on secondary equine embryonic lung monolayers and stored at -70~ This virus stock had a TCID50 of 10663/ml.
Mares Experimentally Infected when 9 Months Pregnant ("Late" Gestation) These animals (nos 6 11), described earlier by Smith et al. (1992, 1993), are included for comparison with those described below (nos 1 5).
Experimental Infection of 3- or 5-Month Pregnant Pony Mares Five Welsh Mountain pony mares (nos 1-5) with known ovulation and insemination dates were purchased from an equine breeding establishment. They were isolated from other equids and monitored serologically to demonstrate stable, low complementfixing antibody titres to EHV-1 and EHV-4 (Thompson et al., 1976). Pregnancy was confirmed at 70 days by assay of pregnant mare serum gonadotrophin. At 3 months of pregnancy ("early" gestation), mares 1 to 4 were individually housed and challenged with EHV-1 Ab4 by intranasal instillation. The inoculum contained 105 (mares 1 and 2) or 106.6 (mares 3 and 4) TCIDs0 of infectious virus in tissue culture supernate, as these doses had been shown to induce high rates of abortion in pony mares infected in late gestation (Mumford et al., 1994). Mare 5 was housed and infected with 105 TCIDs0 of EHV-1 Ab4 at 5 months of pregnancy ("mid"-gestation). Clinical and Hrological Monitoring All five mares were examined clinically each morning and afternoon. In mares 1 - 4 , the afternoon recording of rectal temperature and collection of nasopharyngeal swabs and heparinized blood samples for virus isolation (VI) were carried out as described by Mumford et al. /1994). Because mare 5 was difficult to sample without sedation, virological monitoring of this animal was limited to alternate-day blood sampling. Mares l 5 were humanely destroyed for immediate necropsy and collection of tissue samples on days 8, 9, 11, 12 or 13 post-infection (PI), respectively. The timing of these post-mortem examinations was related to the period of maximal EHV-1 antigen expression in the endometrium of mares infected in late pregnancy with the Ab4 isolate (Smith et al., 1993).
EHV-1
Uterine
Vascular
233
Lesions
Table 1 E H V - 1 i n f e c t i o n o f f o u r m a r e s in e a r l y p r e g n a n c y a n d o n e in m i d - p r e g n a n c y
Mare number
1 2 3 4 5
Stage of gestation (months)
Challenge inoculum (TCID~o)
3 3 3 3 5
l0 S l0 s 1066 l066 105
Intervals (days) after inoculation at which virus recovered Jrom nasopharynx 1,2,4 1,2 2,3,4,5 2,3 ND
viraemia present
animal killed
5,7 5,8 3 9 4 8 --*
8 9 ll 12 13
N D = Not done. * = No viraemia.
Collection and Processing of 7~ssue Samples Ten specimens of endometrium and attached allantochorion (pregnant horn [PH] tip, mid-dorsal PH, PH convexity, mid-dorsal non-pregnant horn [NPH], NPH convexity, dorsal mid-body, ventral mid-body, left mid-body, right mid-body and cervical pole) and samples of the amnion, umbilical cord and major fetal organs were collected into virus transport medium for VI and into neutral buffered formalin for histological examination and immunoperoxidase (IP) labelling as described previously (Smith et al., 1993). After dissection of the uterus, placenta and fetus, specimens of nasal mueosa, turbinate, lung, respiratory tract lymph nodes, liver, spleen, brain and spinal cord were collected and fixed in neutral buffered formalin. The tissues were then processed and embedded in paraffin wax, and sections were stained by haematoxylin and eosin (HE). Viral antigen was "visualized" by indirect IP staining with a rabbit polyclonal EHV-1 antiserum (Whitwell et al., 1992). Tissue specimens in viral transport medium were homogenized manually by grinding with sand and passaged twice on rabbit kidney (RK-13) cell monolayers, with daily examination for cytopathogenic effect. Consecutive paraffin-wax sections of endometrium and placenta from each uterine site were stained with HE and by the IP method, and these sections were used to derive blood vessel counts and percentages. Three counting measurements were obtained from each uterine site: (i) n=total number of endometrial arterioles in section; (ii) r v - n u m b e r of arterioles in section demonstrating vasculitis (fibrinoid necrosis, intramural leucocyte infiltration or marked adventitial leucocyte cuffing); (iii) r~ = number of arterioles in section demonstrating viral antigen expression. These counts were used to obtain the percentages of arterioles demonstrating vasculitis and viral antigen expression at each uterine site, where percentage vasculitis = rv/n x 100, and percentage antigen expression =ra/n x 100, and these percentages were used in subsequent statistical analysis of the results. Mild perivenular or interstitial accumulations of leucocytes were ignored because such changes are also seen in normal pregnant mares (Ginther, 1979). Owing to the relative ease of stripping the allantochorion from the endometrium by dissection in mares in early pregnancy, sites of placental separation seen microscopically were assumed to be artefactual unless an influx of fluid or cells was present at the uteroplacental interface.
Results
Clinical and Virological Findings T h e o u t c o m e o f infection is s u m m a r i z e d in T a b l e 1. All five m a r e s showed signs o f mild u p p e r respiratory tract disease for 1 week after challenge, with
234
K. Q, Smith et M.
low-grade bilateral serous or mucoid nasal discharge, anorexia and dullness. Nasopharyngeal shedding of infectious virus occurred over days 1 to 5 PI, with the development of cell-associated viraemia from day 3 PI onwards. Clinical signs of ataxia, paresis or abortion were absent.
Macroscopical Appearance of the Genital Tract Mares 1 to 4 (3 months gestation; days 8, 9, 11 and 12 PI) were each pregnant with a single fetus in the right uterine horn (average fetal bodyweight 0" 18 kg and average crown-rump length 18 cm). Corpora lutea were present in both ovaries. The endometrial cups appeared as a horseshoe-shaped.band of pale solid plaques at the base of the pregnant horn in mares 1 and 2, with corresponding elliptical avillous sites on the allantochorion. In mares 3 and 4 the cups were partly degenerate, with mucoid honey-like material in the cup depressions, and corresponding allantochorionic pouches were present. No placental separation was recognized in the four mares, and the cervix was closed with an intact mucous seal in each case. There was no evidence of premature m a m m a r y development or relaxation of the sacro-iliac joints. Mare 3 had firm chalky nodules, resembling necrotic fat, in the pelvic canal, and there were old fibrous perivaginal adhesions, presumably reflecting an earlier dystocia. No gross lesions were noted on dissection of the fetuses of mares 1 and 2, but agonal meconium staining and petechiation of the major organs were present in the fetuses of mares 3 and 4. Mare 5 (5 months gestation; day 13 PI) was pregnant in the right uterine horn, which contained a single fetus (bodyweight 1"68 kg) with early growth of the mane, tail and eyelashes. The endometrial cups at the base of the pregnant horn had sloughed, with residual flat fibrous scars at the cup sites. There was no evidence of placental separation, premature m a m m a r y development or relaxation of the sacro-iliac joints, and the cervix was closed, with an intact mucous seal. Dissection of the fetus revealed scattered agonal subcutaneous petechiae only.
~rological and Immunohistological Findings in the Genital Tract At 3 months of gestation (mares 1 to 4), the chorionic villi showed only primary or secondary branching, and interdigitated with the endometrial crypts as a series of undulations (Fig. 1). By 5 months (mare 5), microcotyledonary structures were evident at the uteroplacental interface, but the cotyledons were smaller and had less complex branching than those normally seen in the final third of pregnancy. Mares 1 to 4 (3 months gestation; days 8, 9, 11 and 12 PI) had scattered lymphocytes, haemosiderophages and plasma cells in the endometrial stroma, sometimes forming tiny interstitial, perivenular and periglandular aggregates. Occasional endometrial arterioles in mares 1 and 2 demonstrated mild adventitial cuffing or lymphohistiocytic vasculitis, with viral antigen expression in endothelial cells on "step" (sequential) sectioning and IP staining. Rare early mural non-occlusive fibrin thrombi were noted. Mare 3 demonstrated a
E H V - 1 Uterine V a s c u l a r L e s i o n s
235
Fig. 1.
Normal structure ofmacrovilli showing simple primary and secondalT branching at utero-placental interface in early pregnancy. Mare 1, 3 months gestation, 8 days PI. HE. x 165.
Fig. 2.
Marked periarteriolar lymphohistiocytic cuffing in glandular layer of endometrium. Mare 3, 3 months gestation, 11 days PI. HE. x 330.
patchy marked lymphohistiocytic vasculitis affecting deep endometrial and myometrial arterioles (Fig. 2). Swelling of endothelial cells was noted, and viral antigen was detected in endothelial cells and medial myocytes of affected blood vessels by step sectioning and IP staining (Fig. 3). Focal early thrombosis and fibrinoid mural necrosis were occasionally seen. Vasculitis or fibrinoid mural necrosis was occasionally noted in the endometrial and myometrial arterioles of mare 4, but viral antigen was not detected at the affected sites despite step sectioning and IP staining. All four mares had occasional, probably incidental, periglandular fibrosis and atrophy with inspissation or mineralization of retained secretion.
236
K . C. S m i t h et al.
., e
f
It ~
2
Fig. 3.
Endothelial cell swelling and viral antigen expression in endometrial arteriole. Mare 3, 3 months gestation, 11 days PI. IP. x 660.
Fig. 4.
Marked periarteriolar lymphohistiocytic cuffing and viral antigen expression in tunica media of deep endometrial arteriole. Mare 5, 5 months gestation, 13 days PI. IP. x 330.
EHV-1 U t e r i n e V a s c u l a r L e s i o n s
237
No placental separation was recognized in mares 1-4. Examination of specimens of allantochorion, amnion, umbilical cord and major fetal organs invariably failed to reveaI any significant lesions or foci of viral antigen expression. Specimens of uterine, placental and fetal tissue were virologically negative as shown by two passages on RK13 monolayers. Mare 5 (5 months gestation; day 13 PI) demonstrated frequent lymphohistiocytic vasculitis or adventitial cuffing of arterioles in the subglandular and glandular layers of the endometrium, and in the myometrium. Fibrinoid necrosis was noted as a segmental change, and occasional blood vessels were thrombosed. The endometrium was congested and oedematous, but loci of ischaemic cotyledonary necrosis were not seen. Viral antigen was commonly demonstrated in swollen endothelial cells and medial myocytes of affected arterioles by IP staining (Fig. 4). No placental separations were seen, and no significant lesions or foci of viral antigen expression were recognized in specimens of allantochorion, amnion, umbilical cord or major fetal organs.
Immunohistological Findings Outside the Genital Tract Small erosions with local neutrophil exocytosis were noted in the nasal and turbinate mucosa of mares 1, 2, 3 and 4 (days 8, 9, 11 and 12 PI), and infected endothelial cells and myocytes were present in turbinate arterioles in mares 1, 3 and 4. Infection of endothelial cells and monocytes was noted in lymphatics of the submandibular, retropharyngeal and bronchial lymph nodes in mares 3 and 4; these animals also demonstrated occasional infection of bronchiolar epithelium and pulmonary interstitial endothelium. Infection of small blood vessels in the medulla and spinal cord was present in mares 2 and 5 (days 9 and 13 PI), without local thrombo-ischaemic lesions. A patchy intense cholangiohepatitis was noted incidentally in mare 5, and tiny neutrophil aggregates were present in the liver of mares 1 and 2.
Statistical Comparison of Data for Early versus Late Pregnancy The percentages of endometrial arterioles demonstrating vasculitis or viral antigen expression at each of the 10 uterine sites are given in Table 2. These data, representing estimates of the overall percentages of vasculitis and viral antigen expression in each uterus, were compared with previously published results obtained in an identical manner from six pony mares challenged with EHV-1 Ab4 and examined on days 6, 8, 9, 10, 11 and 13 PI (Smith et al., 1993), as presented in Tables 3 and 4 (mares 6 to 11). This comparison was not simple, since the total number of sampling units (11 mares) was small, and a variety of statistical analyses was used to check that different approaches yielded similar conclusions. The use of the Student's t-test to compare mares challenged at 3 months and 9 months required the two sets of mares to have been examined at the same times PI. Restricting the analysis to data collected on days 8, 9 and 11 PI (i.e. comparing ponies 1, 2, 3 with 7, 8, 10) made the test valid, but reduced
238
K.C.
S m i t h e t al.
Table 2 P e r c e n t a g e s of e n d o m e t r i a l a r t e r i o l e s s h o w i n g v a s c u l i t i s a n d viral a n t i g e n e x p r e s s i o n at 10 u t e r i n e s i t e s in p o n y m a r e s in e a r l y p r e g n a n c y or m i d - p r e g n a n c y % ofendometrial arterioles showing vasculitis/ and viral antigen in mare no.
Ute~ne
site
Tip PH Dorsal PH Convexity PH Dorsal NPH Convexity NPH Dorsal body Ventral body Left body Right body Cervical pole Mean
1"
2*
3*
4*
5~
5/0 9/0 7/0 0/0 5/0 6/0 4/2 0/0 0/0 4/0
0/0 0/0 6/6 0/0 0/0 0/0 0/0 7/0 0/0 0/0
21/0 6/6 6/0 8/5 14/0 40/0 43/0 16/4 20/2 13/0
6/0 11/0 12/0 14/0 0/0 2"5/0 0/0 16/0 0/0 3/0
76/24 40/6 48/14 49/5 65/13 48/12 45/9 29/0 50/7 0/0
4"0/0"2
1"3/0"6
18'7/1"7
6"45/0"0
45'0/9"0
PH = Pregnant horn; NPH = non-pregnant horn. Animals were * 3 months pregnant or "~5 months pregnant. See Materials and Methods for intervals between inoculation and killing.
Table 3 EHV-1 infection of s i x 9 - m o n t h - p r e g n a n t p o n y m a r e s Mare no.
bwculum (TCID @
Days PI on which viraemiapresent
Occurrence of neurological disease (day PI)
Occurrence of abortion (day YI)
Day PI on which animal killed
6 7 8 9 l0
107~ 104 107~ 107 10~
3 6 5-7 3 9 4 8 5 9
No No No Severe (9) Severe (10)
6 8 9 10 11
11
10 4
5 11
Severe (12)
Yes No Yes (9) No Dead fetus in utew No
13
PI = post-infection.
the total sample size to six. The results are summarized in Tables 5 and 6. For both percentage vasculitis and percentage antigen expression, the observed means for the early pregnant mares were smaller than those for the late pregnant mares, so that all t values were negative. However, due to the small sample size, the t-tests failed to reach the conventional significance level of P<0"05. "Principal components analysis" (PCA) examines the structure and reduces the "dimensionality" of multivariate data (Anderson, 1984; Dillon and Goldstein, 1984). It was appropriate here because there was a large number of variables (corresponding to the uterine sites in Tables 2 and 4) for each pony, and PCA could be applied to the data on all the ponies. Separate analyses were carried out for percentage vasculitis and percentage antigen expression. The results were similar for both measurements and only the percentage vasculitis analysis is presented. The first principal component, PC 1, explained
239
EHV-1 Uterine Vascular Lesions
Table 4 P e r c e n t a g e s of e n d o m e t r i a l arterioles s h o w i n g vasculitis and viral antigen e x p r e s s i o n at 10 uterine s i t e s in six 9-month-pregnant mares* Uterine site
Tip PH Mid dorsal PH Convexity PH Mid dorsal NPH Convexity NPH Dorsal midbody Ventral midbody Left midbody Right midbody Caudal body/ cervical pole Mean
% of endometrial arterioles showing vasculitis/ and viral antigen in mare no. 6
7
8
9
10
11
0/0 0/0 0/0 0/0 7/0 0/0 0/0 0/0 0/0 50/36
0/0 0/0 0/0 0/0 0/0 0/0 0/0 0/0 13/0 10/0
68/37 65/24 55/36 73/40 83/39 89/56 45/9 78/44 71/12 54/15
47/17 25/19 29/8 22/6 45/14 30/15 17/0 39/9 25/7 6/0
29/20 42/25 64/43 55/30 62/19 50/0 58/25 65/43 44/15 29/29
57/30 52/14 45/20 40/8 43/14 52/39 13/0 43/14 65/29 18/12
5"7/3"6
2"3/0"0
68'1/31"2
28"5/9'5
49"8/24"9
42"8/18'0
* See Smith et al. (1992, 1993). PH = pregnant horn; NPH = non-pregnant horn. Mares 6 11 were killed, respectively, 6, 8, 9, 10, 11 and 13 days after inoculation.
Table 5 M e a n s of the p e r c e n t a g e s of e n d o m e t r i a l arterioles s h o w i n g vasculitis at different uterine sites Uterine site Tip PH Dorsal PH Convexity PH Dorsal NPH Convexity NPH Dorsal body Ventral body Left body Right body Cervical pole Mean
Mean (3-month)
Mean (9-month)
Student's t-statistic
Significance probability
8"67 5"00 6'33 2"67 6"33 15'33 15'67 7"66 6"67 5"67
32'33 35"67 39"67 42"67 48"33 46'33 34"33 47"67 42'67 31 "00
1"14 - 1"60 1"67 - 1"81 - 1"66 1'08 --0"84 - 1"63 2"00 -- 1"90
0"32 0" 19 0" 17 0" 15 0' 17 0"34 0'44 0" 18 0" 12 0" l 3
8'00
40"07
1"58
0" 19
PH = pregnant horn; NPH = non-pregnant horn. Arithmetic means: 3-month and 9-month pregnant mares examined on days 8, 9 and 11 post-infection, and Student's t-statistic (with 4 d.f.) for testing the equality of means for the early and late pregnant mare populations.
85% of the total variance in the 10 variables, and PC2 explained a further 6% of the total variance: A scatter diagram of PC1 and PC2 values (Fig. 5) was based upon 91% of the total information in the data. This figure shows that the 3-month pregnant mares were clustered with low values of PC 1. In contrast, the 9-month pregnant mares examined on days 9, 10, 11 and 13 PI had high values of PC1. The 9-month pregnant mares examined on days 6 and 8 PI were closer to the 3-month pregnant mares. In the group of 3-month
240
K.C.
Means
of the
percentages
of endometrial
Uterine site
Smith
al.
et
Table 6 arterioles showing uterine sites
antigen
expression
at different
Mean (3-month)
~Viean (9-month)
Student's t statistic
Significance probability
Tip PH Dorsal PH Convexity PH Dorsal NPH Convexity NPH Dorsal body Ventral body Left body Right body Cervical pole
0'00 2"00 2"00 1-67 0'00 0'00 0"67 1-33 0"67 0"00
19'00 16'33 26" 33 23"33 19"33 18"67 11 "33 29'00 9-00 14"67
1.78 - 1'70 -- 1'81 - 1"79 1"72 1"00 - 1-45 - 1.90 1.80 - 1"75
0" 15 0" 16 0" 15 0"15 0"16 0'38 0"22 0" 13 0" 15 0" 15
Mean
0-83
18"70
1.87
0" 13
PH =pregnant horn; NPH = non-pregnant horn. Arithmetic means: 3-month and 9-month pregnant mares examined on days 8, 9 and 11 post-infection, and Student's t-statistic (with 4 d.f.) for testing the equality of means for the early and late pregnant mare populations.
40 f
20 ~q
o-g ba
-20
-40 -8(
'
I
-60
,
I
-40
,
r
-20
,
,
0
I
20
,
I
40
,
I
60
e
I
80
,
I
100
J
I
120
~
r
140
Values of PC1 Fig. 5.
Scatter diagram of PC2 and PC1 values for early, mid and late pregnant mares. The four 3month-pregnant mares 1, 2, 3 and 4 are shown as a, b, c and d; mare 5 (5-month-pregnant) by e; and the six 9-month-pregnant mares 6, 7, 8, 9, 10 and 11 by f, g, h, i, j and k.
pregnant mares, the two animals examined on days 11 and 12 PI were to the right. The 5-month pregnant mare was indistinguishable from the 9-month pregnant group along the PC 1 axis, but along the PC2 axis this mare had the lowest value.
EHV-1 Uterine
Vascular
241
Lesions
80
60 a= >
5 4O
N 20
, 0
2
I 4
,
I 6
,
~ 8
I 10
I 12
, 14
Days post-infection Fig. 6.
Obsmwed increase in m e a n percentage of arterioles showing vasculitis in 9-month-pregnant mares. T h e fitted logistic model has an upper limit K = 4 7 % , and 95% confidence limits of 30-64%.
Modelling the data in time was important as the observed responses reflected a temporal growth process after infection. Many biological processes have a slow initial growth rate, followed by a period of rapid growth, and then a decline in growth rate, with variable values fluctuating around an upper asymptote. The non-linear logistic model mimics such sigmoid growth. If Y = response variable and X = time in days post infection, a suitable logistic model is: Y = K x exp(A+ B x X ) / [ I + exp(A+ B x X)]; in which K, A and B are the parameters of the model. The parameter K, defining the upper asymptotic value of the relationship between Y and X, was of particular interest in the present study. It was difficult to fit the model described above to the available data with its large biological variability and few time points (days 8, 9, 11 and 12 for "3-month" mares; and days 6, 8, 9, 10, 11 and 13 for "9-month" mares), and a further value, Y = 0 when X = 0 , was created; this facilitated model fitting, without affecting the estimate of K. This analysis was restricted to the mean values of percentage vasculitis and percentage antigen expression, for the 9-month pregnant mare data. In Fig. 6, showing the observed increase in the mean percentage of arterioles demonstrating vasculitis in the 9-month pregnant ponies, the estimate of the upper asymptote K was 47%, and the 95% confidence limits for the population asymptote were 30% to 64%. Against these figures, the highest observed mean value for the 3-month mares was 18"7%. The corresponding values for the percentage antigen data were an
242
K.C. Smith et al.
estimate of K = 21% and 95% confidence limits of 11 31% for the 9-month mares, whereas the maximum observed mean value for 3-month mares was 1"7%. Discussion
The data indicated that EHV-1 replication and antigen expression occurred in the endometrium of mares infected intranasally with the Ab4 isolate at 3 months of gestation. The fact that the number of mares employed in the study was small necessitated extra care in the statistical analysis and interpretation of the results. The simple, univariate t-testing procedure could not be used efficiently because the early and late gestational ponies were not all examined at the same times PI. Unlike the t-testing approach, the principal components analysis used all of the observations, while the modelling approach further took into account the biological nature of the observations, which described a growth process in time. The results from these different statistical analyses indicated consistently that the endometrial vascular changes and viral antigen expression in endothelial cells of early pregnant mares were less than in mares infected with EHV-1 in late pregnancy and examined after comparable incubation periods. Since percentages rather than absolute numbers of blood vessels were used in the analyses, no correction was made for the greater overall surface area and vascularity of uteri in late pregnancy than in early pregnancy. Thrombosis was rare and no thrombo-ischaemic lesions were recognized in the mares in early pregnancy. The dearth of thrombotic lesions at this stage of gestation contrasts with the findings 9 to 13 days PI in mares infected at 9 months of pregnancy, when multifocal thrombosis and cotyledonary infarction were associated with placental separation and abortion or fetal death in utero (Smith et al., 1002, 1993). The degree of vasculitis and antigen expression in the endometrium of the single 5-month-pregnant mare was comparable with that occurring in 9-month-pregnant animals, but this may have been due, at least in part, to the fact that this mare was examined relatively late in the incubation period. Since the difference in EHV-1 antigen expression in early versus late pregnancy was relative rather than absolute, with small numbers of vascular lesions occurring in all of the mares in early pregnancy, any resistance of mares to abortigenic EHV-1 infection in the first 3 months of gestation must be related to differences in both the degree and the outcome of viral replication in the endometrium. These differences might be the result of (1) specific immunological and endocrine effects influencing the number of infected arterioles, and (2) non-specific effects stemming from the anatomy of the placental barrier and the mechanisms by which pregnancy is maintained. The theory of a general immunosuppressive effect associated with pregnancy (Medawar, 1953) is controversial. Clearly, complete immunosuppression rendering pregnant females susceptible to repeated opportunistic infections does not occur, but there may be a slight pregnancy-associated depression in cellmediated immunity (Weinberg, 1984). Gerber et al. (1977) suggest that cellmediated responses to EHV- 1 infection, as assessed by lymphocyte stimulation
EHV-1 U t e r i n e V a s c u l a r L e s i o n s
243
assay, are suppressed in pregnancy, but that there is no concurrent effect on humoral immunity. The experimental studies of EHV-I paresis by Jackson et aL (1977) showed that only mares which were pregnant (3, 6, 8 and 9 months of gestation) when infected with EHV-1 Army 183 developed neurological signs, whereas non-pregnant infected mares remained clinically normal. Dalsgaard (1970) made similar field observations, and Mumford et al. (1987) observed a lower incidence of EHV-1 infection in non-pregnant than in pregnant mares during a natural outbreak. However, the resistance of nonpregnant horses, including males, to EHV-1 neurological disease has not been borne out by other reports (Thompson et aL, 1979; Platt et al., 1980; Crowhurst et al., 1981; Thein, 1981; Kohn and Fenner, 1987; Whitwell and Blunden, 1992; Blunden et al., 1993). Given that systemic pregnancy-associated immunosuppression does not completely explain the effect of stage of pregnancy on the degree of EHV-1 replication, the local effect of steroid and gonadotrophic hormones in the uterine microenvironment should be considered. Cortisone, progesterone and oestrogen at high concentration all suppress lymphocyte responsiveness in vitro (Siiteri et al., 1977; Stites and Siiteri, 1983), and progesterone has an immunosuppressive effect both in vivo and in vitro at concentrations existing at the fetomaternal interface in man and various laboratory animals (Beer and Billingham, 1979; Stites and Siiteri, 1983). It is likely that any direct effect of progesterone in facilitating viral infection of the endometrium is greater when the hormone is secreted at the uteroplacental interface in late pregnancy than when released into the circulation from the corpora lutea in the first 4-5 months (Ginther, 1979). Edington et al. (1991) stated that the resistance of the pregnant mare to EHV-1 abortion in early gestation may be related to the hormonal stability of the placenta in response to vascular injury. Platelet aggregation in thrombosis is known to cause the local release of vasoactive substances such as serotonin, thromboxanes and prostaglandins (Taussig, 1984; Cotran et al., 1989; Robinson and Maxie, 1993), and such prostaglandin release as a result of uterine viral thrombosis in late pregnancy may play a role in EHV-1 abortion (Smith, 1994). Since thrombo-ischaemic damage did not occur in our 3-monthpregnant mares, the equine endometrium may be less susceptible to such damage and to the resultant prostaglandin release in early than in late gestation. The role of prostaglandins in equine pregnancy failure due to endotoxaemia prior to day 55 of gestation has been reviewed by Daels et al. (1989). Endotoxin causes sequestration of neutrophils and platelets in capillary beds and consequent prostaglandin release. Prostaglandin-mediated luteolysis then results in early embryonic death and abortion. Mares are refractory to the abortigenic effects of experimentally administered endotoxin after day 55 of gestation (Daels et al., 1989), and abortion is difficult to induce with exogenous prostaglandins from day 40 to 140, during which period circulating pregnant mare serum gonadotrophin (PMSG) concentrations remain high (Allen and Rowson, 1973; Allen, 1975). The rare cases of EHV-1 abortion that occur in the 4th and 5th months of gestation are different from the typical late abortion of a
244
K.C. Smith et al.
live fetus, being associated with expulsion of a grossly autolysed fetus and placenta. This delay in expulsion after fetal death reflects the fact that a viable fetoplacental unit is not required for the maintenance of high circulating concentrations of progesterone from days 40 to 140; this is because the hormone is secreted by the primary and secondary corpora lutea under the influence of PMSG from the endometrial cups, and luteolysis and abortion do not occur until spontaneous cup regression and sloughing take place. Clotting factors increase during h u m a n pregnancy (Brewer and Aubry, 1973; Walker et al., 1994), and Jackson et al. (1977) suggested that a similar change in late equine pregnancy would enhance E H V - l - i n d u c e d thrombosis and neurological disease. This is supported by recent data indicating that plasma fibrinogen, factor VIII:C and Von Willebrand factor activity gradually increase from mid-gestation to parturition in normal mares (Gentry et al., 1992). These increases may contribute, in part, to thrombosis on virus-infected endometrial endothelium, and hence to a greater risk of abortion in the final third of pregnancy. The discrete vascular supply of the mature microcotyledon is established from day 150, and provides for efficient transfer of gases by functioning as a countercurrent exchange mechanism (Steven, 1968; Samuel et al., 1974, 1976); this also results in an effective end-circulation at each microcotyledon. Cotyledonary infarction following thrombosis in the endometrium is, therefore, facilitated in late gestation. The slight vascular changes seen in our small group of 3-month-pregnant mares are unlikely to have facilitated either transplacental infection or abortion, but the eventual outcome of these pregnancies had the animals been allowed to survive cannot be known. Because EHV-1 can become latent in lymphoreticular tissues and trigeminal ganglia after acute infection (Welch et al., 1992; Edington et al., 1994; Slater et al., 1994), the possibility of latent infection in early pregnancy and subsequent reactivation as the endometrium becomes more susceptible to viral damage in late pregnancy is not discounted. EHV1 latency and reactivation during pregnancy, or prolonged low-level viraemia, may account for the wide variation in abortion incubation periods which in field outbreaks can vary from 9 days to several months (Mumford et al., 1987). The cellular sites of EHV-1 latency are likely to be identified in the near future by the application of sensitive molecular techniques such as in-situ hybridization for viral latency-associated transcripts (Edington, 1992) in conjunction with the novel in-situ polymerase chain reaction for low copy number viral DNA (Gressens and Martin, 1994). There will then be a need for experimental EHV-1 infection of further groups of mares in early pregnancy and intensive clinical and virological monitoring throughout gestation in order to detect the early establishment of latency and possible viral reactivation and abortion towards full term.
Acknowledgments The authors are grateful to Professor N. Edington and Miss Katherine E. Whitwell for valuable discussion and comments on these data. Assistance with immunoperoxidase staining was provided by Mrs S. M. Gower, and virus isolation from nasopharyngeal
EHVol Uterine Vascular L e s i o n s
245
swabs and blood samples was made by Ms T.-A. Hammond and Ms Z. Swann. K.C.S. was in receipt of a Wellcome Trust Veterinary Research Training Scholarship. References
Allen, G. P. and Bryans, J. T. (1986). Molecular epizootiology, pathogenesis and prophylaxis of equine herpesvirus- 1 infections. In: Progress in VeterinaryMicrobiology and Immunology, Vol. 2, Karger, Basel, pp. 78 144. Allen, W. E. (1975). Pregnancy failure induced by human chorionic gonadotrophin in pony mares. Veterinary Record, 96, 88-90. Allen, W. R. and Rowson, L. E. (1973). Control of the mare's oestrous cycle by prostaglandins. Journal of Reproduction and Fertility, Suppl., 33, 539-543. Anderson, T. W. (1984). In: An Introduction to Multivariate StatisticalAnalysis, 2nd Edit., John Wiley, New York, p. 675. Beer, A. E. and Billingham, R. E. (1979). Maternal immunological recognition mechanisms during pregnancy. In: Maternal Recognition of Pregnancy, Ciba Series 64, Excerpta Medica, Oxford, pp. 293-308. Blunden, A. S., Whitwell, K. E. and Pegler, K. M. (1993). An outbreak of paralysis associated with equine herpesvirus type 1 infection in a livery stable. Progress in Veterinary Neurology, 3, 95 100. Brewer, D. W. and Aubry, R. H. (1973). The physiology of pregnancy: clinical pathological correlations. Postgraduate Medicine, 53, 221-226. Bryans, J. T. (1981). Application of management procedures and prophylactic immunisation to the control of equine rhinopneumonitis. In: Proceedings of the 26th Annual Congress of the American Association of Equine Practitioners, 26, pp. 259-272. Cotran, R. S., Kumar, V. and Robbins, S. R. (1989). Fluid and haemodynamic derangements. In: Robbins' Pathologic Basis of Disease, W. B. Saunders Company, Philadelphia, pp. 87 120. Crowhurst, F. A., Dickinson, G. and Burrows, R. (1981). An outbreak of paresis in mares and geldings associated with equid herpesvirus-1. Veterinary Record, 109, 527-528. Daels, P. F., Stabenfeldt, G. H., Kindahl, H. and Hughes, J. P. (1989). Prostaglandin release and luteolysis associated with physiological and pathological conditions of the reproductive tract of the mare: a review. Equine VeterinaryJournal, Suppl., 8, 29 34. Dalsgaard, H. (1970). Enzootic paresis in horses in association with an outbreak of rhinopneumonitis (virus abortion). Medlemsbl Danske Drylaegeforen, 53, 71-76. Dillon, W. R. and Goldstein, M. (1984). In: Multivariate Analysis, John Wiley, New York, p. 587. Doll, E. R. (1952). Seasonal incidence and fetal age in equine virus abortion. Cornell Veterinarian, 42, 505-509. Doll, E. R. and Bryans, J. T. (1963). Epizootiology of equine viral rhinopneumonitis. Journal of the American Veterinary Medical Association, 142, 31-37. Doll, E. R., Crowe, M. E. W., Bryans, J. T. and McCollum, W. H. (1955). Infection immunity in equine virus abortion. Cornell Veterinarian, 55, 387 410. Edington, N. (1992). Latency of equine herpesviruses. In: Proceedingsof the 6th International Conference on Equine Infectious Diseases, Cambridge 1991, W. Plowright, P. D. Rossdale andJ. F. Wade, Eds, R&W Publications, Newmarket, pp. 195-200. Edington, N., Smyth, B. and Griffiths, L. (1991). The role of endothelial cell infection in the endometrium, placenta and fetus in equid herpesvirus-1 abortions. Journal of Comparative Pathology, 104, 379-387. Edington, N., Welch, H. M. and Griffiths, L. (1994). The prevalence of latent equid herpesviruses in the tissues of 40 abattoir horses. Equine Veterinary Journal, 26, 140-142. Gentry, P. A., Feldman, B. F., O'Neill, S. L., Madigan, J. E. and Zinkl, J. G. (1992). Evaluation of the haemostatic profile in the pre- and post-parturient mare, with
246
K.C. Smith et al.
particular focus on the perinatal period. Equine Veterinary Journal, 24, 33 36. Gerber, J. D., Marron, A. D., Bass, E. P. and Bechenhaurer, W. H. (1977). Effect of age and pregnancy on the antibody and cell-mediated immune responses of horses to equine herpesvirus-1. Canadian Journal of Comparative Medicine, 41, 471-478. Ginther, O.J. (1979). Reproductive Biology of the Mare: Basic and Applied Aspects, Equiservices, Cross Plaines, Wisconsin. Gressens, P. and Martin, J. R. (1994). In situ polymerase chain reaction: localisation of HSV-2 DNA sequences in infections of the nervous system. Journal of Virological Methods, 46, 61 83. Jackson, T. A., Osburn, B. I., Cordy, D. R. and Kendrick, J. W. (1977). Equine herpesvirus-1 infection of horses: studies on the experimentally-induced neurologic disease. Journal of the American Veterinary Medical Association, 38, 709 719. Kohn, C. W. and Fenner, W. R. (1987). Equine herpes myeloencephalopathy. Veterinary Clinics of North America (Equine Practice), 3, 405-419. Medawar, P. B. (1953). Some immunological and endocrinological problems raised by the evolution of viviparity in vertebrates. Symposia of the Society of Experimental Biology, VII, 320-338. Mumford, J. A., Hannant, D.,Jessett, D. M., Smith, K. C. and Ostlund, E. N. (1994). Abortigenic and neurological disease caused by equid herpesvirus-1. In: Proceedings of the 7th International Conference on Equine Infectious Diseases, Tokyo 1994, W. Plowright and H. Nakajima, Eds, R&W Publications, Newmarket, pp. 261-275. Mumford, J. A., Rossdale, P. D., Jessett, D. M., Gann, S.J., Ousey, J. and Cook, R. F. (1987). Serological and virological investigations of an equid herpesvirus1 (EHV-1) abortion storm on a stud farm in 1985. Journal of Reproduction and Fertility, Suppl., 35, 509 518. Platt, H., Singh, H. and Whitwell, K. E. (1980). Pathological observations on an outbreak of paralysis in broodmares. Equine VeterinaryJournal, 12, 118-126. Prickett, M. E. (1970). The pathology of disease caused by equine herpesvirus-1. In: Proceedings of the 2nd International Conference on Equine Infectious Diseases, Paris 1969, J. T. Bryans and H. Gerber, Eds, S. Karger AG, Basel, pp. 24 33. Robinson, W. F. and Maxie, J. G. (1993). The cardiovascular system. In: Pathology of Domestic Animals, Vol. 3, 4th Edit., K. V. F.Jubb, P. C. Kennedy and N. Palmer, Eds, Academic Press, San Diego, pp. 1-100. Samuel, C. A., Allen, W. R. and Steven, D. H. (1974). Studies on the equine placenta. I: Development of the microcotyledons. Journal of Reproduction and Fertility, 41, 441-445. Samuel, C. A., Allen, W. R. and Steven, D. H. (1976). Studies on the equine placenta. II: Ultrastructure of the placental barrier. Journal of Reproduction and Fertility, 48, 257 264. Siiteri, P. K., Febrs, F., Clemens, L. E., Chang, R.J., Gondos, B. and Stites, D. P. (1977). Progesterone and maintenance of pregnancy: is progesterone nature's immunosuppressant? Annals of the New York Academy of Sciences, 286, 384-387. Slater, J. D., Borchers, K., Thackray, A. M. and Field, H.J. (1994). The trigeminal ganglion is a location for equine herpesvirus-1 (EHV~ 1) latency and reactivation in the horse. Journal of General Virology, 75, 2007-2016. Smith, K. C. (1994). Studies on the pathogenesis of abortigenic infection by equid herpesvirus-1 in pregnant pony mares. PhD thesis, University of London. Smith, K. C., Whitwell, K. E., Binns, M. M., Dolby, C. A., Hannant, D. and Mumford, J. A. (1992). Abortion of virologically negative fetuses following experimental challenge of pregnant pony mares with equid herpesvirus-1. Equine Veterinary Journal, 24, 256-259. Smith, K. C., Whitwell, K. E., Mumford, J. A., Gower, S. M., Hannant, D. and Tearle, J. P. (1993). An immunohistological study of the uterus of mares following experimental infection with equid herpesvirus-1. Equine Veterinary Journal, 25, 36-40.
EHV-1 Uterine V a s c u l a r L e s i o n s
247
Steven, D. H. (1968). Structural differences between exchange units in the sheep and horse. Journal of Physiology, 196, 24-26. Stites, D. P. and Siiteri, P. K. (1983). Steroids as immunosuppressants in pregnancy. Immunology Reviews, 75, 117-139. Taussig, M.J. (1984). The circulation. In: Processes in Pathology and Microbiology, 2nd Edit., Blackwell Scientific Publications, Oxford, pp. 623-688. Thein, P. (1981). Infection of the central nervous system with equine herpesvirus serotype 1. Journal of the South African VeterinaryAssociation, 52, 239-241. Thompson, G. R., Mumford, J. A., Campbell, J., Griffith, L. and Clapham, P. (1976). Serological detection of equid herpesvirus-1 infections in the respiratory tract. Equine VeterinaryJournal, 8, 58-65. Thompson, G. W., McCready, R., Sanford, E. and Gagnon, A. (1979). An outbreak of herpesvirus myeloencephalitis in vaccinated horses. Canadian VeterinaryJournal, 20, 22-25. Walker, I. D., Walker, J. j., Colvin, B. T., Letsky, E. A., Rivers, R. and Stevens, R. (1994). Investigation and management of haemorrhagic disorders in pregnancy. Journal of Clinical Pathology, 47, 100-108. Weinberg, E. D. (1984). Pregnancy associated depression of cell-mediated immunity. Reviews of Infectious Diseases, 6, 814-827. Welch, H. M., Bridges, C. G., Lyon, A. M., Griffiths, L. and Edington, N. (1992). Latent equid herpesviruses 1 and 4: detection and distinction using the polymerase chain reaction and co-cultivation from lymphoid tissues. Journal of General ~rology, 73, 261-268. Whitwell, K. E. and Blunden, A. S. (1992). Pathological findings in horses dying during an outbreak of the paralytic form of equid herpesvirus type 1 (EHV-1) infection. Equine VeterinaryJournal, 24, 13-19. Whitwell, K. E., Gower, S. M. and Smith, K. C. (1992). An immunoperoxidase method applied to the diagnosis of equine herpesvirus abortion, using conventional and rapid microwave techniques. Equine VeterinaryJournal, 24, 10-12.
7
A Received,June 2nd, 1995 ccepted, November 3rd, 1995J
A Comparison of Equid Herpesvirus-1 (EHV-1) Vascular Lesions in the Early versus Late Pregnant Equine Uterus K. t3. Smith, J. A. M u m f o r d and K. Lakhani Animal Health Trust, Centrefor Preventive Medicine, P O. Box 5, Newmarket, Suffolk CB8 7DW,, UK
Summary Four Welsh Mountain pony mares at 3 months of gestation and one mare at 5 months were inoculated intranasally with equid herpesvirus-1 (EHV-I: Ab4 isolate) at doses of 105 to 106.6 TCID50. All five mares became infected, but no cases of paresis or abortion occurred. On days 8, 9, 1 l, 12 (3-monthpregnant mares) and 13 (5-month-pregnant mare) after infection, a detailed examination of the pregnant uterus was made. Small numbers of vascular lesions with EHV-I antigen expression in endothelial cells were present in the uteri of the early gestational mares; thrombi were rare and loci ofthromboischaemic damage were not seen. Six pony mares previously inoculated with EHV-1 Ab4 at 9 months of gestation had a significantly greater degree of vascular abnormality than that found in the four mares infected at 3 months of gestation, but the degree of EHV-1 antigen expression and thrombosis in the uterus was similar to that found in the single mare infected when 5 months pregnant. 9 1996 W.B. Saunders Company Limited
Introduction Equid herpesvirus-1 (EHV-1) is an 0~-herpesvirus which remains a common abortigenic pathogen in Equidae despite widespread use of vaccination and management procedures designed to limit the spread of infection in susceptible populations (Allen and Bryans, 1986). Experimental studies with the highly virulent EHV-1 isolates Army 183 and Ab4, which also cause neurological disease, have demonstrated the role of endothelial cell infection and thrombosis in the endometrium of the late pregnant uterus in producing abortion (Edington et al., 1991; Smith et al., 1992, 1993). Field studies have shown that 95% of EHV-1 abortions occur in the last 4 months of gestation (Doll, 1952; Doll and Bryans, 1963). This apparent resistance of the early pregnant mare to abortigenic EHV-1 infection formed the basis of a programme of abortion control practised on studfarms in Kentucky for many years (Doll and Bryans, 1963). The programme was based on the intranasal administration of a live hamster-adapted EHV-1 isolate to broodmares in early pregnancy in order to provoke a moderate immune response with a low risk of vaccine-induced abortion. A booster inoculation was given 4 months later in order to reinforce immunity during the critical final 0021 9975/96/030231 + 17 $12.00/0
9 1996 W.B. Saunders Company Limited
232
K.C. Smith et al.
third o f p r e g n a n c y . This control p r o g r a m m e posed f o r m i d a b l e m a n a g e m e n t p r o b l e m s but r e d u c e d the incidence o f a b o r t i o n epizootics w h e n rigorously applied in successive b r e e d i n g seasons (Bryans, 1981). E H V - 1 a b o r t i o n has b e e n r e c o r d e d on rare occasions as early as the f o u r t h m o n t h o f gestation (Prickett, 1970) and, after e x p e r i m e n t a l infection, at 3 m o n t h s (Doll et al., 1955). T o investigate the low susceptibility o f the p r e g n a n t m a r e to abortigenic E H V - 1 infection in early gestation, four mares at 3 m o n t h s o f p r e g n a n c y a n d one m a r e at 5 m o n t h s were infected intranasally with the Ab4 isolate o f E H V 1, m o n i t o r e d clinically a n d virologically, a n d n e c r o p s i e d on days 8 to 13 after inoculation. T h e results were c o m p a r e d with those arising f r o m earlier experiments in which mares in late p r e g n a n c y were infected with the Ab4 isolate (Smith et al., 1992, 1993). Materials and Methods
[/grus
The Ab4 isolate of EHV-1 was obtained from a quadriplegic mare affected during an outbreak of neurological disease in 1980 (Crowhurst et al., 1981). The virus was passaged eight times on secondary equine embryonic lung monolayers and stored at -70~ This virus stock had a TCID50 of 10663/ml.
Mares Experimentally Infected when 9 Months Pregnant ("Late" Gestation) These animals (nos 6 11), described earlier by Smith et al. (1992, 1993), are included for comparison with those described below (nos 1 5).
Experimental Infection of 3- or 5-Month Pregnant Pony Mares Five Welsh Mountain pony mares (nos 1-5) with known ovulation and insemination dates were purchased from an equine breeding establishment. They were isolated from other equids and monitored serologically to demonstrate stable, low complementfixing antibody titres to EHV-1 and EHV-4 (Thompson et al., 1976). Pregnancy was confirmed at 70 days by assay of pregnant mare serum gonadotrophin. At 3 months of pregnancy ("early" gestation), mares 1 to 4 were individually housed and challenged with EHV-1 Ab4 by intranasal instillation. The inoculum contained 105 (mares 1 and 2) or 106.6 (mares 3 and 4) TCIDs0 of infectious virus in tissue culture supernate, as these doses had been shown to induce high rates of abortion in pony mares infected in late gestation (Mumford et al., 1994). Mare 5 was housed and infected with 105 TCIDs0 of EHV-1 Ab4 at 5 months of pregnancy ("mid"-gestation). Clinical and Hrological Monitoring All five mares were examined clinically each morning and afternoon. In mares 1 - 4 , the afternoon recording of rectal temperature and collection of nasopharyngeal swabs and heparinized blood samples for virus isolation (VI) were carried out as described by Mumford et al. /1994). Because mare 5 was difficult to sample without sedation, virological monitoring of this animal was limited to alternate-day blood sampling. Mares l 5 were humanely destroyed for immediate necropsy and collection of tissue samples on days 8, 9, 11, 12 or 13 post-infection (PI), respectively. The timing of these post-mortem examinations was related to the period of maximal EHV-1 antigen expression in the endometrium of mares infected in late pregnancy with the Ab4 isolate (Smith et al., 1993).
EHV-1
Uterine
Vascular
233
Lesions
Table 1 E H V - 1 i n f e c t i o n o f f o u r m a r e s in e a r l y p r e g n a n c y a n d o n e in m i d - p r e g n a n c y
Mare number
1 2 3 4 5
Stage of gestation (months)
Challenge inoculum (TCID~o)
3 3 3 3 5
l0 S l0 s 1066 l066 105
Intervals (days) after inoculation at which virus recovered Jrom nasopharynx 1,2,4 1,2 2,3,4,5 2,3 ND
viraemia present
animal killed
5,7 5,8 3 9 4 8 --*
8 9 ll 12 13
N D = Not done. * = No viraemia.
Collection and Processing of 7~ssue Samples Ten specimens of endometrium and attached allantochorion (pregnant horn [PH] tip, mid-dorsal PH, PH convexity, mid-dorsal non-pregnant horn [NPH], NPH convexity, dorsal mid-body, ventral mid-body, left mid-body, right mid-body and cervical pole) and samples of the amnion, umbilical cord and major fetal organs were collected into virus transport medium for VI and into neutral buffered formalin for histological examination and immunoperoxidase (IP) labelling as described previously (Smith et al., 1993). After dissection of the uterus, placenta and fetus, specimens of nasal mueosa, turbinate, lung, respiratory tract lymph nodes, liver, spleen, brain and spinal cord were collected and fixed in neutral buffered formalin. The tissues were then processed and embedded in paraffin wax, and sections were stained by haematoxylin and eosin (HE). Viral antigen was "visualized" by indirect IP staining with a rabbit polyclonal EHV-1 antiserum (Whitwell et al., 1992). Tissue specimens in viral transport medium were homogenized manually by grinding with sand and passaged twice on rabbit kidney (RK-13) cell monolayers, with daily examination for cytopathogenic effect. Consecutive paraffin-wax sections of endometrium and placenta from each uterine site were stained with HE and by the IP method, and these sections were used to derive blood vessel counts and percentages. Three counting measurements were obtained from each uterine site: (i) n=total number of endometrial arterioles in section; (ii) r v - n u m b e r of arterioles in section demonstrating vasculitis (fibrinoid necrosis, intramural leucocyte infiltration or marked adventitial leucocyte cuffing); (iii) r~ = number of arterioles in section demonstrating viral antigen expression. These counts were used to obtain the percentages of arterioles demonstrating vasculitis and viral antigen expression at each uterine site, where percentage vasculitis = rv/n x 100, and percentage antigen expression =ra/n x 100, and these percentages were used in subsequent statistical analysis of the results. Mild perivenular or interstitial accumulations of leucocytes were ignored because such changes are also seen in normal pregnant mares (Ginther, 1979). Owing to the relative ease of stripping the allantochorion from the endometrium by dissection in mares in early pregnancy, sites of placental separation seen microscopically were assumed to be artefactual unless an influx of fluid or cells was present at the uteroplacental interface.
Results
Clinical and Virological Findings T h e o u t c o m e o f infection is s u m m a r i z e d in T a b l e 1. All five m a r e s showed signs o f mild u p p e r respiratory tract disease for 1 week after challenge, with
234
K. Q, Smith et M.
low-grade bilateral serous or mucoid nasal discharge, anorexia and dullness. Nasopharyngeal shedding of infectious virus occurred over days 1 to 5 PI, with the development of cell-associated viraemia from day 3 PI onwards. Clinical signs of ataxia, paresis or abortion were absent.
Macroscopical Appearance of the Genital Tract Mares 1 to 4 (3 months gestation; days 8, 9, 11 and 12 PI) were each pregnant with a single fetus in the right uterine horn (average fetal bodyweight 0" 18 kg and average crown-rump length 18 cm). Corpora lutea were present in both ovaries. The endometrial cups appeared as a horseshoe-shaped.band of pale solid plaques at the base of the pregnant horn in mares 1 and 2, with corresponding elliptical avillous sites on the allantochorion. In mares 3 and 4 the cups were partly degenerate, with mucoid honey-like material in the cup depressions, and corresponding allantochorionic pouches were present. No placental separation was recognized in the four mares, and the cervix was closed with an intact mucous seal in each case. There was no evidence of premature m a m m a r y development or relaxation of the sacro-iliac joints. Mare 3 had firm chalky nodules, resembling necrotic fat, in the pelvic canal, and there were old fibrous perivaginal adhesions, presumably reflecting an earlier dystocia. No gross lesions were noted on dissection of the fetuses of mares 1 and 2, but agonal meconium staining and petechiation of the major organs were present in the fetuses of mares 3 and 4. Mare 5 (5 months gestation; day 13 PI) was pregnant in the right uterine horn, which contained a single fetus (bodyweight 1"68 kg) with early growth of the mane, tail and eyelashes. The endometrial cups at the base of the pregnant horn had sloughed, with residual flat fibrous scars at the cup sites. There was no evidence of placental separation, premature m a m m a r y development or relaxation of the sacro-iliac joints, and the cervix was closed, with an intact mucous seal. Dissection of the fetus revealed scattered agonal subcutaneous petechiae only.
~rological and Immunohistological Findings in the Genital Tract At 3 months of gestation (mares 1 to 4), the chorionic villi showed only primary or secondary branching, and interdigitated with the endometrial crypts as a series of undulations (Fig. 1). By 5 months (mare 5), microcotyledonary structures were evident at the uteroplacental interface, but the cotyledons were smaller and had less complex branching than those normally seen in the final third of pregnancy. Mares 1 to 4 (3 months gestation; days 8, 9, 11 and 12 PI) had scattered lymphocytes, haemosiderophages and plasma cells in the endometrial stroma, sometimes forming tiny interstitial, perivenular and periglandular aggregates. Occasional endometrial arterioles in mares 1 and 2 demonstrated mild adventitial cuffing or lymphohistiocytic vasculitis, with viral antigen expression in endothelial cells on "step" (sequential) sectioning and IP staining. Rare early mural non-occlusive fibrin thrombi were noted. Mare 3 demonstrated a
E H V - 1 Uterine V a s c u l a r L e s i o n s
235
Fig. 1.
Normal structure ofmacrovilli showing simple primary and secondalT branching at utero-placental interface in early pregnancy. Mare 1, 3 months gestation, 8 days PI. HE. x 165.
Fig. 2.
Marked periarteriolar lymphohistiocytic cuffing in glandular layer of endometrium. Mare 3, 3 months gestation, 11 days PI. HE. x 330.
patchy marked lymphohistiocytic vasculitis affecting deep endometrial and myometrial arterioles (Fig. 2). Swelling of endothelial cells was noted, and viral antigen was detected in endothelial cells and medial myocytes of affected blood vessels by step sectioning and IP staining (Fig. 3). Focal early thrombosis and fibrinoid mural necrosis were occasionally seen. Vasculitis or fibrinoid mural necrosis was occasionally noted in the endometrial and myometrial arterioles of mare 4, but viral antigen was not detected at the affected sites despite step sectioning and IP staining. All four mares had occasional, probably incidental, periglandular fibrosis and atrophy with inspissation or mineralization of retained secretion.
236
K . C. S m i t h et al.
., e
f
It ~
2
Fig. 3.
Endothelial cell swelling and viral antigen expression in endometrial arteriole. Mare 3, 3 months gestation, 11 days PI. IP. x 660.
Fig. 4.
Marked periarteriolar lymphohistiocytic cuffing and viral antigen expression in tunica media of deep endometrial arteriole. Mare 5, 5 months gestation, 13 days PI. IP. x 330.
EHV-1 U t e r i n e V a s c u l a r L e s i o n s
237
No placental separation was recognized in mares 1-4. Examination of specimens of allantochorion, amnion, umbilical cord and major fetal organs invariably failed to reveaI any significant lesions or foci of viral antigen expression. Specimens of uterine, placental and fetal tissue were virologically negative as shown by two passages on RK13 monolayers. Mare 5 (5 months gestation; day 13 PI) demonstrated frequent lymphohistiocytic vasculitis or adventitial cuffing of arterioles in the subglandular and glandular layers of the endometrium, and in the myometrium. Fibrinoid necrosis was noted as a segmental change, and occasional blood vessels were thrombosed. The endometrium was congested and oedematous, but loci of ischaemic cotyledonary necrosis were not seen. Viral antigen was commonly demonstrated in swollen endothelial cells and medial myocytes of affected arterioles by IP staining (Fig. 4). No placental separations were seen, and no significant lesions or foci of viral antigen expression were recognized in specimens of allantochorion, amnion, umbilical cord or major fetal organs.
Immunohistological Findings Outside the Genital Tract Small erosions with local neutrophil exocytosis were noted in the nasal and turbinate mucosa of mares 1, 2, 3 and 4 (days 8, 9, 11 and 12 PI), and infected endothelial cells and myocytes were present in turbinate arterioles in mares 1, 3 and 4. Infection of endothelial cells and monocytes was noted in lymphatics of the submandibular, retropharyngeal and bronchial lymph nodes in mares 3 and 4; these animals also demonstrated occasional infection of bronchiolar epithelium and pulmonary interstitial endothelium. Infection of small blood vessels in the medulla and spinal cord was present in mares 2 and 5 (days 9 and 13 PI), without local thrombo-ischaemic lesions. A patchy intense cholangiohepatitis was noted incidentally in mare 5, and tiny neutrophil aggregates were present in the liver of mares 1 and 2.
Statistical Comparison of Data for Early versus Late Pregnancy The percentages of endometrial arterioles demonstrating vasculitis or viral antigen expression at each of the 10 uterine sites are given in Table 2. These data, representing estimates of the overall percentages of vasculitis and viral antigen expression in each uterus, were compared with previously published results obtained in an identical manner from six pony mares challenged with EHV-1 Ab4 and examined on days 6, 8, 9, 10, 11 and 13 PI (Smith et al., 1993), as presented in Tables 3 and 4 (mares 6 to 11). This comparison was not simple, since the total number of sampling units (11 mares) was small, and a variety of statistical analyses was used to check that different approaches yielded similar conclusions. The use of the Student's t-test to compare mares challenged at 3 months and 9 months required the two sets of mares to have been examined at the same times PI. Restricting the analysis to data collected on days 8, 9 and 11 PI (i.e. comparing ponies 1, 2, 3 with 7, 8, 10) made the test valid, but reduced
238
K.C.
S m i t h e t al.
Table 2 P e r c e n t a g e s of e n d o m e t r i a l a r t e r i o l e s s h o w i n g v a s c u l i t i s a n d viral a n t i g e n e x p r e s s i o n at 10 u t e r i n e s i t e s in p o n y m a r e s in e a r l y p r e g n a n c y or m i d - p r e g n a n c y % ofendometrial arterioles showing vasculitis/ and viral antigen in mare no.
Ute~ne
site
Tip PH Dorsal PH Convexity PH Dorsal NPH Convexity NPH Dorsal body Ventral body Left body Right body Cervical pole Mean
1"
2*
3*
4*
5~
5/0 9/0 7/0 0/0 5/0 6/0 4/2 0/0 0/0 4/0
0/0 0/0 6/6 0/0 0/0 0/0 0/0 7/0 0/0 0/0
21/0 6/6 6/0 8/5 14/0 40/0 43/0 16/4 20/2 13/0
6/0 11/0 12/0 14/0 0/0 2"5/0 0/0 16/0 0/0 3/0
76/24 40/6 48/14 49/5 65/13 48/12 45/9 29/0 50/7 0/0
4"0/0"2
1"3/0"6
18'7/1"7
6"45/0"0
45'0/9"0
PH = Pregnant horn; NPH = non-pregnant horn. Animals were * 3 months pregnant or "~5 months pregnant. See Materials and Methods for intervals between inoculation and killing.
Table 3 EHV-1 infection of s i x 9 - m o n t h - p r e g n a n t p o n y m a r e s Mare no.
bwculum (TCID @
Days PI on which viraemiapresent
Occurrence of neurological disease (day PI)
Occurrence of abortion (day YI)
Day PI on which animal killed
6 7 8 9 l0
107~ 104 107~ 107 10~
3 6 5-7 3 9 4 8 5 9
No No No Severe (9) Severe (10)
6 8 9 10 11
11
10 4
5 11
Severe (12)
Yes No Yes (9) No Dead fetus in utew No
13
PI = post-infection.
the total sample size to six. The results are summarized in Tables 5 and 6. For both percentage vasculitis and percentage antigen expression, the observed means for the early pregnant mares were smaller than those for the late pregnant mares, so that all t values were negative. However, due to the small sample size, the t-tests failed to reach the conventional significance level of P<0"05. "Principal components analysis" (PCA) examines the structure and reduces the "dimensionality" of multivariate data (Anderson, 1984; Dillon and Goldstein, 1984). It was appropriate here because there was a large number of variables (corresponding to the uterine sites in Tables 2 and 4) for each pony, and PCA could be applied to the data on all the ponies. Separate analyses were carried out for percentage vasculitis and percentage antigen expression. The results were similar for both measurements and only the percentage vasculitis analysis is presented. The first principal component, PC 1, explained
239
EHV-1 Uterine Vascular Lesions
Table 4 P e r c e n t a g e s of e n d o m e t r i a l arterioles s h o w i n g vasculitis and viral antigen e x p r e s s i o n at 10 uterine s i t e s in six 9-month-pregnant mares* Uterine site
Tip PH Mid dorsal PH Convexity PH Mid dorsal NPH Convexity NPH Dorsal midbody Ventral midbody Left midbody Right midbody Caudal body/ cervical pole Mean
% of endometrial arterioles showing vasculitis/ and viral antigen in mare no. 6
7
8
9
10
11
0/0 0/0 0/0 0/0 7/0 0/0 0/0 0/0 0/0 50/36
0/0 0/0 0/0 0/0 0/0 0/0 0/0 0/0 13/0 10/0
68/37 65/24 55/36 73/40 83/39 89/56 45/9 78/44 71/12 54/15
47/17 25/19 29/8 22/6 45/14 30/15 17/0 39/9 25/7 6/0
29/20 42/25 64/43 55/30 62/19 50/0 58/25 65/43 44/15 29/29
57/30 52/14 45/20 40/8 43/14 52/39 13/0 43/14 65/29 18/12
5"7/3"6
2"3/0"0
68'1/31"2
28"5/9'5
49"8/24"9
42"8/18'0
* See Smith et al. (1992, 1993). PH = pregnant horn; NPH = non-pregnant horn. Mares 6 11 were killed, respectively, 6, 8, 9, 10, 11 and 13 days after inoculation.
Table 5 M e a n s of the p e r c e n t a g e s of e n d o m e t r i a l arterioles s h o w i n g vasculitis at different uterine sites Uterine site Tip PH Dorsal PH Convexity PH Dorsal NPH Convexity NPH Dorsal body Ventral body Left body Right body Cervical pole Mean
Mean (3-month)
Mean (9-month)
Student's t-statistic
Significance probability
8"67 5"00 6'33 2"67 6"33 15'33 15'67 7"66 6"67 5"67
32'33 35"67 39"67 42"67 48"33 46'33 34"33 47"67 42'67 31 "00
1"14 - 1"60 1"67 - 1"81 - 1"66 1'08 --0"84 - 1"63 2"00 -- 1"90
0"32 0" 19 0" 17 0" 15 0' 17 0"34 0'44 0" 18 0" 12 0" l 3
8'00
40"07
1"58
0" 19
PH = pregnant horn; NPH = non-pregnant horn. Arithmetic means: 3-month and 9-month pregnant mares examined on days 8, 9 and 11 post-infection, and Student's t-statistic (with 4 d.f.) for testing the equality of means for the early and late pregnant mare populations.
85% of the total variance in the 10 variables, and PC2 explained a further 6% of the total variance: A scatter diagram of PC1 and PC2 values (Fig. 5) was based upon 91% of the total information in the data. This figure shows that the 3-month pregnant mares were clustered with low values of PC 1. In contrast, the 9-month pregnant mares examined on days 9, 10, 11 and 13 PI had high values of PC1. The 9-month pregnant mares examined on days 6 and 8 PI were closer to the 3-month pregnant mares. In the group of 3-month
240
K.C.
Means
of the
percentages
of endometrial
Uterine site
Smith
al.
et
Table 6 arterioles showing uterine sites
antigen
expression
at different
Mean (3-month)
~Viean (9-month)
Student's t statistic
Significance probability
Tip PH Dorsal PH Convexity PH Dorsal NPH Convexity NPH Dorsal body Ventral body Left body Right body Cervical pole
0'00 2"00 2"00 1-67 0'00 0'00 0"67 1-33 0"67 0"00
19'00 16'33 26" 33 23"33 19"33 18"67 11 "33 29'00 9-00 14"67
1.78 - 1'70 -- 1'81 - 1"79 1"72 1"00 - 1-45 - 1.90 1.80 - 1"75
0" 15 0" 16 0" 15 0"15 0"16 0'38 0"22 0" 13 0" 15 0" 15
Mean
0-83
18"70
1.87
0" 13
PH =pregnant horn; NPH = non-pregnant horn. Arithmetic means: 3-month and 9-month pregnant mares examined on days 8, 9 and 11 post-infection, and Student's t-statistic (with 4 d.f.) for testing the equality of means for the early and late pregnant mare populations.
40 f
20 ~q
o-g ba
-20
-40 -8(
'
I
-60
,
I
-40
,
r
-20
,
,
0
I
20
,
I
40
,
I
60
e
I
80
,
I
100
J
I
120
~
r
140
Values of PC1 Fig. 5.
Scatter diagram of PC2 and PC1 values for early, mid and late pregnant mares. The four 3month-pregnant mares 1, 2, 3 and 4 are shown as a, b, c and d; mare 5 (5-month-pregnant) by e; and the six 9-month-pregnant mares 6, 7, 8, 9, 10 and 11 by f, g, h, i, j and k.
pregnant mares, the two animals examined on days 11 and 12 PI were to the right. The 5-month pregnant mare was indistinguishable from the 9-month pregnant group along the PC 1 axis, but along the PC2 axis this mare had the lowest value.
EHV-1 Uterine
Vascular
241
Lesions
80
60 a= >
5 4O
N 20
, 0
2
I 4
,
I 6
,
~ 8
I 10
I 12
, 14
Days post-infection Fig. 6.
Obsmwed increase in m e a n percentage of arterioles showing vasculitis in 9-month-pregnant mares. T h e fitted logistic model has an upper limit K = 4 7 % , and 95% confidence limits of 30-64%.
Modelling the data in time was important as the observed responses reflected a temporal growth process after infection. Many biological processes have a slow initial growth rate, followed by a period of rapid growth, and then a decline in growth rate, with variable values fluctuating around an upper asymptote. The non-linear logistic model mimics such sigmoid growth. If Y = response variable and X = time in days post infection, a suitable logistic model is: Y = K x exp(A+ B x X ) / [ I + exp(A+ B x X)]; in which K, A and B are the parameters of the model. The parameter K, defining the upper asymptotic value of the relationship between Y and X, was of particular interest in the present study. It was difficult to fit the model described above to the available data with its large biological variability and few time points (days 8, 9, 11 and 12 for "3-month" mares; and days 6, 8, 9, 10, 11 and 13 for "9-month" mares), and a further value, Y = 0 when X = 0 , was created; this facilitated model fitting, without affecting the estimate of K. This analysis was restricted to the mean values of percentage vasculitis and percentage antigen expression, for the 9-month pregnant mare data. In Fig. 6, showing the observed increase in the mean percentage of arterioles demonstrating vasculitis in the 9-month pregnant ponies, the estimate of the upper asymptote K was 47%, and the 95% confidence limits for the population asymptote were 30% to 64%. Against these figures, the highest observed mean value for the 3-month mares was 18"7%. The corresponding values for the percentage antigen data were an
242
K.C. Smith et al.
estimate of K = 21% and 95% confidence limits of 11 31% for the 9-month mares, whereas the maximum observed mean value for 3-month mares was 1"7%. Discussion
The data indicated that EHV-1 replication and antigen expression occurred in the endometrium of mares infected intranasally with the Ab4 isolate at 3 months of gestation. The fact that the number of mares employed in the study was small necessitated extra care in the statistical analysis and interpretation of the results. The simple, univariate t-testing procedure could not be used efficiently because the early and late gestational ponies were not all examined at the same times PI. Unlike the t-testing approach, the principal components analysis used all of the observations, while the modelling approach further took into account the biological nature of the observations, which described a growth process in time. The results from these different statistical analyses indicated consistently that the endometrial vascular changes and viral antigen expression in endothelial cells of early pregnant mares were less than in mares infected with EHV-1 in late pregnancy and examined after comparable incubation periods. Since percentages rather than absolute numbers of blood vessels were used in the analyses, no correction was made for the greater overall surface area and vascularity of uteri in late pregnancy than in early pregnancy. Thrombosis was rare and no thrombo-ischaemic lesions were recognized in the mares in early pregnancy. The dearth of thrombotic lesions at this stage of gestation contrasts with the findings 9 to 13 days PI in mares infected at 9 months of pregnancy, when multifocal thrombosis and cotyledonary infarction were associated with placental separation and abortion or fetal death in utero (Smith et al., 1002, 1993). The degree of vasculitis and antigen expression in the endometrium of the single 5-month-pregnant mare was comparable with that occurring in 9-month-pregnant animals, but this may have been due, at least in part, to the fact that this mare was examined relatively late in the incubation period. Since the difference in EHV-1 antigen expression in early versus late pregnancy was relative rather than absolute, with small numbers of vascular lesions occurring in all of the mares in early pregnancy, any resistance of mares to abortigenic EHV-1 infection in the first 3 months of gestation must be related to differences in both the degree and the outcome of viral replication in the endometrium. These differences might be the result of (1) specific immunological and endocrine effects influencing the number of infected arterioles, and (2) non-specific effects stemming from the anatomy of the placental barrier and the mechanisms by which pregnancy is maintained. The theory of a general immunosuppressive effect associated with pregnancy (Medawar, 1953) is controversial. Clearly, complete immunosuppression rendering pregnant females susceptible to repeated opportunistic infections does not occur, but there may be a slight pregnancy-associated depression in cellmediated immunity (Weinberg, 1984). Gerber et al. (1977) suggest that cellmediated responses to EHV- 1 infection, as assessed by lymphocyte stimulation
EHV-1 U t e r i n e V a s c u l a r L e s i o n s
243
assay, are suppressed in pregnancy, but that there is no concurrent effect on humoral immunity. The experimental studies of EHV-I paresis by Jackson et aL (1977) showed that only mares which were pregnant (3, 6, 8 and 9 months of gestation) when infected with EHV-1 Army 183 developed neurological signs, whereas non-pregnant infected mares remained clinically normal. Dalsgaard (1970) made similar field observations, and Mumford et al. (1987) observed a lower incidence of EHV-1 infection in non-pregnant than in pregnant mares during a natural outbreak. However, the resistance of nonpregnant horses, including males, to EHV-1 neurological disease has not been borne out by other reports (Thompson et aL, 1979; Platt et al., 1980; Crowhurst et al., 1981; Thein, 1981; Kohn and Fenner, 1987; Whitwell and Blunden, 1992; Blunden et al., 1993). Given that systemic pregnancy-associated immunosuppression does not completely explain the effect of stage of pregnancy on the degree of EHV-1 replication, the local effect of steroid and gonadotrophic hormones in the uterine microenvironment should be considered. Cortisone, progesterone and oestrogen at high concentration all suppress lymphocyte responsiveness in vitro (Siiteri et al., 1977; Stites and Siiteri, 1983), and progesterone has an immunosuppressive effect both in vivo and in vitro at concentrations existing at the fetomaternal interface in man and various laboratory animals (Beer and Billingham, 1979; Stites and Siiteri, 1983). It is likely that any direct effect of progesterone in facilitating viral infection of the endometrium is greater when the hormone is secreted at the uteroplacental interface in late pregnancy than when released into the circulation from the corpora lutea in the first 4-5 months (Ginther, 1979). Edington et al. (1991) stated that the resistance of the pregnant mare to EHV-1 abortion in early gestation may be related to the hormonal stability of the placenta in response to vascular injury. Platelet aggregation in thrombosis is known to cause the local release of vasoactive substances such as serotonin, thromboxanes and prostaglandins (Taussig, 1984; Cotran et al., 1989; Robinson and Maxie, 1993), and such prostaglandin release as a result of uterine viral thrombosis in late pregnancy may play a role in EHV-1 abortion (Smith, 1994). Since thrombo-ischaemic damage did not occur in our 3-monthpregnant mares, the equine endometrium may be less susceptible to such damage and to the resultant prostaglandin release in early than in late gestation. The role of prostaglandins in equine pregnancy failure due to endotoxaemia prior to day 55 of gestation has been reviewed by Daels et al. (1989). Endotoxin causes sequestration of neutrophils and platelets in capillary beds and consequent prostaglandin release. Prostaglandin-mediated luteolysis then results in early embryonic death and abortion. Mares are refractory to the abortigenic effects of experimentally administered endotoxin after day 55 of gestation (Daels et al., 1989), and abortion is difficult to induce with exogenous prostaglandins from day 40 to 140, during which period circulating pregnant mare serum gonadotrophin (PMSG) concentrations remain high (Allen and Rowson, 1973; Allen, 1975). The rare cases of EHV-1 abortion that occur in the 4th and 5th months of gestation are different from the typical late abortion of a
244
K.C. Smith et al.
live fetus, being associated with expulsion of a grossly autolysed fetus and placenta. This delay in expulsion after fetal death reflects the fact that a viable fetoplacental unit is not required for the maintenance of high circulating concentrations of progesterone from days 40 to 140; this is because the hormone is secreted by the primary and secondary corpora lutea under the influence of PMSG from the endometrial cups, and luteolysis and abortion do not occur until spontaneous cup regression and sloughing take place. Clotting factors increase during h u m a n pregnancy (Brewer and Aubry, 1973; Walker et al., 1994), and Jackson et al. (1977) suggested that a similar change in late equine pregnancy would enhance E H V - l - i n d u c e d thrombosis and neurological disease. This is supported by recent data indicating that plasma fibrinogen, factor VIII:C and Von Willebrand factor activity gradually increase from mid-gestation to parturition in normal mares (Gentry et al., 1992). These increases may contribute, in part, to thrombosis on virus-infected endometrial endothelium, and hence to a greater risk of abortion in the final third of pregnancy. The discrete vascular supply of the mature microcotyledon is established from day 150, and provides for efficient transfer of gases by functioning as a countercurrent exchange mechanism (Steven, 1968; Samuel et al., 1974, 1976); this also results in an effective end-circulation at each microcotyledon. Cotyledonary infarction following thrombosis in the endometrium is, therefore, facilitated in late gestation. The slight vascular changes seen in our small group of 3-month-pregnant mares are unlikely to have facilitated either transplacental infection or abortion, but the eventual outcome of these pregnancies had the animals been allowed to survive cannot be known. Because EHV-1 can become latent in lymphoreticular tissues and trigeminal ganglia after acute infection (Welch et al., 1992; Edington et al., 1994; Slater et al., 1994), the possibility of latent infection in early pregnancy and subsequent reactivation as the endometrium becomes more susceptible to viral damage in late pregnancy is not discounted. EHV1 latency and reactivation during pregnancy, or prolonged low-level viraemia, may account for the wide variation in abortion incubation periods which in field outbreaks can vary from 9 days to several months (Mumford et al., 1987). The cellular sites of EHV-1 latency are likely to be identified in the near future by the application of sensitive molecular techniques such as in-situ hybridization for viral latency-associated transcripts (Edington, 1992) in conjunction with the novel in-situ polymerase chain reaction for low copy number viral DNA (Gressens and Martin, 1994). There will then be a need for experimental EHV-1 infection of further groups of mares in early pregnancy and intensive clinical and virological monitoring throughout gestation in order to detect the early establishment of latency and possible viral reactivation and abortion towards full term.
Acknowledgments The authors are grateful to Professor N. Edington and Miss Katherine E. Whitwell for valuable discussion and comments on these data. Assistance with immunoperoxidase staining was provided by Mrs S. M. Gower, and virus isolation from nasopharyngeal
EHVol Uterine Vascular L e s i o n s
245
swabs and blood samples was made by Ms T.-A. Hammond and Ms Z. Swann. K.C.S. was in receipt of a Wellcome Trust Veterinary Research Training Scholarship. References
Allen, G. P. and Bryans, J. T. (1986). Molecular epizootiology, pathogenesis and prophylaxis of equine herpesvirus- 1 infections. In: Progress in VeterinaryMicrobiology and Immunology, Vol. 2, Karger, Basel, pp. 78 144. Allen, W. E. (1975). Pregnancy failure induced by human chorionic gonadotrophin in pony mares. Veterinary Record, 96, 88-90. Allen, W. R. and Rowson, L. E. (1973). Control of the mare's oestrous cycle by prostaglandins. Journal of Reproduction and Fertility, Suppl., 33, 539-543. Anderson, T. W. (1984). In: An Introduction to Multivariate StatisticalAnalysis, 2nd Edit., John Wiley, New York, p. 675. Beer, A. E. and Billingham, R. E. (1979). Maternal immunological recognition mechanisms during pregnancy. In: Maternal Recognition of Pregnancy, Ciba Series 64, Excerpta Medica, Oxford, pp. 293-308. Blunden, A. S., Whitwell, K. E. and Pegler, K. M. (1993). An outbreak of paralysis associated with equine herpesvirus type 1 infection in a livery stable. Progress in Veterinary Neurology, 3, 95 100. Brewer, D. W. and Aubry, R. H. (1973). The physiology of pregnancy: clinical pathological correlations. Postgraduate Medicine, 53, 221-226. Bryans, J. T. (1981). Application of management procedures and prophylactic immunisation to the control of equine rhinopneumonitis. In: Proceedings of the 26th Annual Congress of the American Association of Equine Practitioners, 26, pp. 259-272. Cotran, R. S., Kumar, V. and Robbins, S. R. (1989). Fluid and haemodynamic derangements. In: Robbins' Pathologic Basis of Disease, W. B. Saunders Company, Philadelphia, pp. 87 120. Crowhurst, F. A., Dickinson, G. and Burrows, R. (1981). An outbreak of paresis in mares and geldings associated with equid herpesvirus-1. Veterinary Record, 109, 527-528. Daels, P. F., Stabenfeldt, G. H., Kindahl, H. and Hughes, J. P. (1989). Prostaglandin release and luteolysis associated with physiological and pathological conditions of the reproductive tract of the mare: a review. Equine VeterinaryJournal, Suppl., 8, 29 34. Dalsgaard, H. (1970). Enzootic paresis in horses in association with an outbreak of rhinopneumonitis (virus abortion). Medlemsbl Danske Drylaegeforen, 53, 71-76. Dillon, W. R. and Goldstein, M. (1984). In: Multivariate Analysis, John Wiley, New York, p. 587. Doll, E. R. (1952). Seasonal incidence and fetal age in equine virus abortion. Cornell Veterinarian, 42, 505-509. Doll, E. R. and Bryans, J. T. (1963). Epizootiology of equine viral rhinopneumonitis. Journal of the American Veterinary Medical Association, 142, 31-37. Doll, E. R., Crowe, M. E. W., Bryans, J. T. and McCollum, W. H. (1955). Infection immunity in equine virus abortion. Cornell Veterinarian, 55, 387 410. Edington, N. (1992). Latency of equine herpesviruses. In: Proceedingsof the 6th International Conference on Equine Infectious Diseases, Cambridge 1991, W. Plowright, P. D. Rossdale andJ. F. Wade, Eds, R&W Publications, Newmarket, pp. 195-200. Edington, N., Smyth, B. and Griffiths, L. (1991). The role of endothelial cell infection in the endometrium, placenta and fetus in equid herpesvirus-1 abortions. Journal of Comparative Pathology, 104, 379-387. Edington, N., Welch, H. M. and Griffiths, L. (1994). The prevalence of latent equid herpesviruses in the tissues of 40 abattoir horses. Equine Veterinary Journal, 26, 140-142. Gentry, P. A., Feldman, B. F., O'Neill, S. L., Madigan, J. E. and Zinkl, J. G. (1992). Evaluation of the haemostatic profile in the pre- and post-parturient mare, with
246
K.C. Smith et al.
particular focus on the perinatal period. Equine Veterinary Journal, 24, 33 36. Gerber, J. D., Marron, A. D., Bass, E. P. and Bechenhaurer, W. H. (1977). Effect of age and pregnancy on the antibody and cell-mediated immune responses of horses to equine herpesvirus-1. Canadian Journal of Comparative Medicine, 41, 471-478. Ginther, O.J. (1979). Reproductive Biology of the Mare: Basic and Applied Aspects, Equiservices, Cross Plaines, Wisconsin. Gressens, P. and Martin, J. R. (1994). In situ polymerase chain reaction: localisation of HSV-2 DNA sequences in infections of the nervous system. Journal of Virological Methods, 46, 61 83. Jackson, T. A., Osburn, B. I., Cordy, D. R. and Kendrick, J. W. (1977). Equine herpesvirus-1 infection of horses: studies on the experimentally-induced neurologic disease. Journal of the American Veterinary Medical Association, 38, 709 719. Kohn, C. W. and Fenner, W. R. (1987). Equine herpes myeloencephalopathy. Veterinary Clinics of North America (Equine Practice), 3, 405-419. Medawar, P. B. (1953). Some immunological and endocrinological problems raised by the evolution of viviparity in vertebrates. Symposia of the Society of Experimental Biology, VII, 320-338. Mumford, J. A., Hannant, D.,Jessett, D. M., Smith, K. C. and Ostlund, E. N. (1994). Abortigenic and neurological disease caused by equid herpesvirus-1. In: Proceedings of the 7th International Conference on Equine Infectious Diseases, Tokyo 1994, W. Plowright and H. Nakajima, Eds, R&W Publications, Newmarket, pp. 261-275. Mumford, J. A., Rossdale, P. D., Jessett, D. M., Gann, S.J., Ousey, J. and Cook, R. F. (1987). Serological and virological investigations of an equid herpesvirus1 (EHV-1) abortion storm on a stud farm in 1985. Journal of Reproduction and Fertility, Suppl., 35, 509 518. Platt, H., Singh, H. and Whitwell, K. E. (1980). Pathological observations on an outbreak of paralysis in broodmares. Equine VeterinaryJournal, 12, 118-126. Prickett, M. E. (1970). The pathology of disease caused by equine herpesvirus-1. In: Proceedings of the 2nd International Conference on Equine Infectious Diseases, Paris 1969, J. T. Bryans and H. Gerber, Eds, S. Karger AG, Basel, pp. 24 33. Robinson, W. F. and Maxie, J. G. (1993). The cardiovascular system. In: Pathology of Domestic Animals, Vol. 3, 4th Edit., K. V. F.Jubb, P. C. Kennedy and N. Palmer, Eds, Academic Press, San Diego, pp. 1-100. Samuel, C. A., Allen, W. R. and Steven, D. H. (1974). Studies on the equine placenta. I: Development of the microcotyledons. Journal of Reproduction and Fertility, 41, 441-445. Samuel, C. A., Allen, W. R. and Steven, D. H. (1976). Studies on the equine placenta. II: Ultrastructure of the placental barrier. Journal of Reproduction and Fertility, 48, 257 264. Siiteri, P. K., Febrs, F., Clemens, L. E., Chang, R.J., Gondos, B. and Stites, D. P. (1977). Progesterone and maintenance of pregnancy: is progesterone nature's immunosuppressant? Annals of the New York Academy of Sciences, 286, 384-387. Slater, J. D., Borchers, K., Thackray, A. M. and Field, H.J. (1994). The trigeminal ganglion is a location for equine herpesvirus-1 (EHV~ 1) latency and reactivation in the horse. Journal of General Virology, 75, 2007-2016. Smith, K. C. (1994). Studies on the pathogenesis of abortigenic infection by equid herpesvirus-1 in pregnant pony mares. PhD thesis, University of London. Smith, K. C., Whitwell, K. E., Binns, M. M., Dolby, C. A., Hannant, D. and Mumford, J. A. (1992). Abortion of virologically negative fetuses following experimental challenge of pregnant pony mares with equid herpesvirus-1. Equine Veterinary Journal, 24, 256-259. Smith, K. C., Whitwell, K. E., Mumford, J. A., Gower, S. M., Hannant, D. and Tearle, J. P. (1993). An immunohistological study of the uterus of mares following experimental infection with equid herpesvirus-1. Equine Veterinary Journal, 25, 36-40.
EHV-1 Uterine V a s c u l a r L e s i o n s
247
Steven, D. H. (1968). Structural differences between exchange units in the sheep and horse. Journal of Physiology, 196, 24-26. Stites, D. P. and Siiteri, P. K. (1983). Steroids as immunosuppressants in pregnancy. Immunology Reviews, 75, 117-139. Taussig, M.J. (1984). The circulation. In: Processes in Pathology and Microbiology, 2nd Edit., Blackwell Scientific Publications, Oxford, pp. 623-688. Thein, P. (1981). Infection of the central nervous system with equine herpesvirus serotype 1. Journal of the South African VeterinaryAssociation, 52, 239-241. Thompson, G. R., Mumford, J. A., Campbell, J., Griffith, L. and Clapham, P. (1976). Serological detection of equid herpesvirus-1 infections in the respiratory tract. Equine VeterinaryJournal, 8, 58-65. Thompson, G. W., McCready, R., Sanford, E. and Gagnon, A. (1979). An outbreak of herpesvirus myeloencephalitis in vaccinated horses. Canadian VeterinaryJournal, 20, 22-25. Walker, I. D., Walker, J. j., Colvin, B. T., Letsky, E. A., Rivers, R. and Stevens, R. (1994). Investigation and management of haemorrhagic disorders in pregnancy. Journal of Clinical Pathology, 47, 100-108. Weinberg, E. D. (1984). Pregnancy associated depression of cell-mediated immunity. Reviews of Infectious Diseases, 6, 814-827. Welch, H. M., Bridges, C. G., Lyon, A. M., Griffiths, L. and Edington, N. (1992). Latent equid herpesviruses 1 and 4: detection and distinction using the polymerase chain reaction and co-cultivation from lymphoid tissues. Journal of General ~rology, 73, 261-268. Whitwell, K. E. and Blunden, A. S. (1992). Pathological findings in horses dying during an outbreak of the paralytic form of equid herpesvirus type 1 (EHV-1) infection. Equine VeterinaryJournal, 24, 13-19. Whitwell, K. E., Gower, S. M. and Smith, K. C. (1992). An immunoperoxidase method applied to the diagnosis of equine herpesvirus abortion, using conventional and rapid microwave techniques. Equine VeterinaryJournal, 24, 10-12.
7
A Received,June 2nd, 1995 ccepted, November 3rd, 1995J