Materials and Methods

Construction of VHH1 expression vectors The immunization of llama with rhesus-monkey rotavirus serotype G3, strain RRV, and subsequent generation and selection of the VHH1 fragment have been described elsewhere [20]. For the generation of the surface-expressed antibody fragment, the DNA fragment coding for VHH1–E-tag gene (E-tagged for detection and purification) was fused to an anchor sequence (the last 244 aa of the proteinase P protein of Lactobacillus casei) into the Lactobacillus expression vector pLP501 cut with restriction enzymes ClaI and XhoI [15, 21]. To generate the secreted VHH1 antibody fragment, a stop codon (TAA) was inserted by polymerase chain reaction (PCR) amplification after the E-tag, and the product was introduced into pLP501. Transformation of L. paracasei (previously named L. casei ATCC 393 pLZ15⁻) [22] was performed as described elsewhere [14]. In both constructs (figure 1), the expression is under the control of the constitutive ldh promoter of L. casei [21]. Lactobacillus expressing an irrelevant VHH-secreted fragment (directed against a Lactococcus phage protein) and an irrelevant VHH-anchored fragment (directed against the SA I/II adhesin of S. mutans) were constructed in a similar way and were used as controls

Bacterial culture The transformed lactobacilli were cultured in Mann-Rogosa Sharp broth that contained 3 μg/mL erythromycin to an OD₆₀₀ of 0.8 (1×10⁸ cfu/mL). Nontransformed L. paracasei were grown without erythromycin

Purification of antibody fragments secreted by lactobacilli Purification of secreted VHH1 and irrelevant VHH antibody fragments was performed using the RPAS Purification module (Amersham Biosciences) [15]. The purity of antibody fragments was verified on SDS-PAGE, and the concentration of total protein was determined using the Bio-Rad protein assay (Bio-Rad Laboratories)

ELISA and flow cytometry Ninety-six–well ELISA plates were coated with rabbit anti–human rotavirus antiserum (1:1000), followed by incubation with RRV (1:100). The cell homogenates or supernatants of Lactobacillus transformants were prepared as described elsewhere [15] and added to the plates. Plates were then incubated with mouse anti–E-tag antibodies (1:1000; Amersham Pharmacia Biotech), followed by alkaline phosphatase (AP)–conjugated goat anti-mouse antibodies (DAKO A/S) (1:1000). VHH1–E-tag purified from the supernatant of lactobacillus secreting VHH1–E-tag was used to determine the concentration of antibody fragments produced by the different transformants. Biotinylated yeast-produced VHH1 followed by incubation with AP-conjugated streptavidin was also used as a positive control

Flow cytometry was performed in accordance with standard protocols using anti–E-tag antibodies followed by cy2-labeled donkey anti-mouse antibody (Jackson Immunoresearch Laboratories), and the samples were analyzed using a FACSCalibur machine (Becton Dickinson) [15]

Scanning electron microscopy (SEM) Cultures of lactobacilli expressing VHH1 fragments anchored to the surface and the nontransformed L. paracasei were mixed with RRV, fixed in 2% glutaraldehyde, and analyzed by SEM (JEOL JSM-820; Jeol) at 15 kV

Virus production and purification RRV was grown and harvested from MA104 cells, and viral titers were determined as described elsewhere [23]. A single virus stock was produced for the entire study

In vitro neutralization assay Lactobacilli expressing anchored VHH fragments (both VHH1 and irrelevant) or affinity-purified VHH1 and irrelevant VHH fragments produced by lactobacilli were serially diluted and incubated for 1 h at room temperature with 200 focus-forming units (ffu) of trypsin-activated RRV (200 μL) before the subsequent inoculation of MA104 cell monolayers [10, 23]. A reduction in the number of RRV-infected cells by >60% relative to the number in control wells was taken to suggest significant neutralization [10]

In vivo assays All mouse experiments were approved by the local ethics committee of the Karolinska Institutet at Karolinska University Hospital, Huddinge, Sweden. Pregnant BALB/c mice were purchased from Møllegard Breeding Center. Four-day-old pups were used for the study. Lactobacilli were administered to pups once daily in a 10-μL volume, starting on day −1 and continuing until day 3. Infections were induced orally on day 0 using 2×10⁷ ffu of RRV (20 diarrhea doses [DD₅₀]), a high dose that was used in the original description of this mouse model [24] and that causes diarrhea in >90% of inoculated mice. In another set of experiments, 4×10⁶ ffu RRV (4 DD₅₀) was used to induce infection

The occurrence of diarrhea was recorded daily until day 4 [10]. Pups were euthanized on day 4, and sections of small intestine were stabilized in RNAlater (Qiagen) for RNA isolation and were fixed in neutral buffered formalin for histopathological analysis. Four independent experiments were conducted: initial testing of the dose-response behavior of the lactobacilli producing anchored VHH1 fragments and subsequent testing of the transformants at the optimal dose. Lactobacilli expressing irrelevant anchored VHH antibody fragment (against the S. mutans SA I/II adhesin) or nontransformed lactobacilli were included in each experiment. An untreated group (RRV-infected only) was also included

To evaluate the survival of lactobacilli in the intestine of mice, pups were fed lactobacilli expressing anchored VHH1 once on day −1, and one-half of them were infected with RRV on day 0. Two pups in each group were euthanized on days 1, 3, 7, and 14, and extracts from sections of small intestine were cultured on Rogosa plates that contained 3 μg/mL erythromycin. PCR was used for detection of the VHH1 insert in erythromycin-resistant colonies

In situ expression of surface-anchored VHH1 on lactobacilli isolated from feces On day 4 after infection, intestinal-wash samples from mice fed daily with lactobacilli expressing anchored VHH were smeared on glass slides (Super Frost Plus; Menzel-Gläser) and were fixed with methanol:acetone (1:1). Slides were successively incubated with mouse anti–E-tag antibody and a cy2-labeled donkey anti-mouse antibody (Jackson Immunoresearch) and were examined using a fluorescence microscope. Controls included fresh cultures of lactobacilli expressing anchored VHH fragments and intestinal washes from naive mice spiked with recombinant lactobacilli or unspiked

Histological analysis Sections of small intestines were stained with hematoxylin-eosin in accordance with standard protocols, and individual slides were evaluated blindly for typical signs of rotavirus infection [24]. Two or 3 sections from 3–4 mice were analyzed in each group

Quantification of RNA Total cellular RNA was isolated from small-intestine tissue, treated with RNase-free DNase (Qiagen), and analyzed by real-time PCR using the EZ RT-PCR core reagent kit (PE Applied Biosystems). Rotavirus vp7 mRNA or viral genomic RNA was amplified at 58°C (ABI 7000 cycler; Applied Biosystems) in the presence of 600 nmol/L primers, 300 nmol/L probe, and 5 mmol/L Mn, to generate a 121-bp–long amplicon. The sense primer (VP7 forward, 5′-CCAAGGGAAAATGTAGCAGTAATTC-3′; nt 791–815), the antisense primer (VP7 reverse, 5′-TGCCACCATTCTTTCCAATTAA-3′; nt 891–912), and the probe (5′-6FAM-TAACGGCTGATCCAACCACAGCACC-TAMRA-3′; nt 843–867) were designed on the basis of the vp7 gene sequence of RRV (GenBank accession number AF295303). A standard curve was generated using a plasmid that contained an RRV vp7 gene, and the lowest level of detection of the PCR was 10 viral RNA copies. The RNA samples from each mouse were normalized against the GAPDH gene [25]. The presence of <10 copies of vp7 RNA was defined as clearance of infection

Statistical analysis Diarrhea in the pups was assessed on the basis of consistency of feces. Watery diarrhea was given a score of 2, loose stool a score of 1, and no stool or normal stool a score of 0. The presence or absence of diarrhea was compared among treatment groups in a daywise manner using Fisher’s exact test and was presented as the percentage of diarrhea in graphs. Severity was defined as the sum of diarrhea scores for each pup during the course of the experiment (severity=Σ diarrhea score [day 1+day 2+day 3+day 4]), and duration was defined as the total sum of days with diarrhea. Both severity and duration were analyzed using Kruskal-Wallis and Dunn tests. Differences in the intestinal virus load as assessed by real-time PCR were tested using the Mann-Whitney U test

Results

Expression and binding of VHH1 antibody fragments ex pressed by recombinant lactobacilli Lactobacilli expressing anchored and secreted VHHI directed against rotavirus were constructed (figure 1). Lactobacilli expressing VHH1 anchored on the surface were referred to as “VHH1-anchored lactobacilli,” and lactobacilli expressing secreted VHH1 as “VHH1-secreted lactobacilli.” Surface expression of VHH1 by L. paracasei was demonstrated by flow cytometry using an anti–E-tag antibody (figure 2A )

Functional activity—that is, binding of rotavirus by the antibody fragments—was first analyzed using ELISA. Binding of antibody fragments to rotavirus was observed when homogenates of transformed VHH1-anchored lactobacilli and supernatant from the VHH1-secreted lactobacilli were used. When purified VHH1–E-tag was used as a standard, the VHH1-anchored lactobacilli were found to express ∼1×10⁴ VHH fragments/bacterium. Supernatants from VHH1-secreted lactobacilli, grown to an OD₆₀₀ of 0.8, contained ∼1 μg/mL of antibody fragments, and the production rate of secreted VHH1 fragments was estimated to be 1×10⁵ VHH1 fragments/bacterium/h. No activity was observed with supernatants or cell extracts from nontransformed L. paracasei or transformed lactobacilli expressing and secreting an irrelevant VHH (against a Lactococcus phage protein) or expressing an irrelevant anchored VHH (against the SA I/II adhesin of S. mutans)

Next, lactobacilli expressing anchored VHH1 were mixed with rotavirus and analyzed by SEM. A large number of virus particles bound to recombinant lactobacilli but not to nontransformed lactobacilli (figure 2B and 2C )

Neutralization of rotavirus in vitro Affinity-purified VHH1 from the supernatant of lactobacilli secreting VHH1 could also neutralize virus in a dose-dependent manner, starting at 60 ng/mL. At 125 ng/mL, the reduction of infectivity was 80% relative to a control VHH (VHH purified from the supernatant of lactobacilli secreting VHH against a Lactococcus phage protein) (figure 3A ). VHH1-anchored lactobacilli could effectively neutralize rotavirus in a dose-dependent manner, starting at 1000 bacteria (figure 3B )

Survival and in situ expression of VHH1-anchored lactobacilli in the mouse gastrointestinal tract After the administration of a single oral dose of VHH1-anchored lactobacilli (1×10⁸ cfu), the transformed bacteria were detected in jejunum and ileum 48 h after treatment, as judged by bacterial culture from intestinal extracts. No difference in the level of cultivable VHH1-anchored lactobacilli was observed between the rotavirus-infected and -uninfected groups. No transformants were detected 96 h after treatment

In situ expression of VHH1 by lactobacilli in the intestinal wash of treated mice was shown by indirect fluorescence using anti–E-tag antibodies. Fluorescent rod-shaped bacteria were observed in intestinal washes from mice fed with VHH1-anchored lactobacilli and in intestinal washes from naive mice that had been spiked with VHH1-anchored lactobacilli but not in unspiked intestinal washes from naive mice. A fresh bacterial culture from lactobacilli expressing VHH1 was used as positive control, to confirm the presence of lactobacilli (data not shown)

Protection by recombinant lactobacilli in a mouse pup model of rotavirus infection The prophylactic effect of transformed lactobacilli was tested in the mouse pup model using a single oral inoculum of 20 DD₅₀ of RRV on day 0. Lactobacilli were administered to pups once daily, starting on day −1 and continuing until day 3. On the basis of the results obtained by testing different doses of lactobacilli for diarrhea intervention, 1×10⁸ cfu was chosen as the optimal dose (figure 4A )

The prevalence of diarrhea was markedly lower in mice treated with VHH1-anchored lactobacilli, compared with that in untreated mice (day 2, P<.002; day 3, P<.0001) or that in mice treated with irrelevant VHH-anchored lactobacilli (day 3, P<.03) (figure 4). In addition, VHH1-anchored lactobacilli significantly shortened the duration (by 0.9 days) and the severity score (by ∼50%), compared with that in untreated mice (P<.01 and P<.001, respectively) or those treated with irrelevant VHH-anchored lactobacilli (P<.05 and P<.01) (table 1). Reconstituted freeze-dried VHH1-anchored lactobacilli were shown to be equally as protective as their fresh counterparts in reducing both diarrhea prevalence (day 2 vs. untreated, P=.03) (data not shown), duration, and severity (vs. untreated, P<.05) (table 1). Interestingly, the VHH1-secreting lactobacilli did not offer better protection than nontransformed lactobacilli (figure 4B ; table 1). The latter induced a modest, statistically nonsignificant (P>.05) reduction in diarrhea prevalence (day 3), duration (by 0.2 days), and severity (by 23%) (figure 4B ; table 1)

Histological examination of the proximal small intestine (jejunum) on day 4 revealed a prevention of inflammation in rotavirus-infected mice treated with VHH1-anchored lactobacilli; in some mice, the histological characteristics were entirely normal. Mice that had received nontransformed lactobacilli showed a mild reduction in pathological changes, whereas mice that received no lactobacilli showed typical signs of rotavirus infection, with swelling and vacuolization of the tips of the villi (not stainable because of epithelial cell death) (figure 5A–5D )

For analysis of the infectious load in the small intestine, expression of rotavirus vp7 was analyzed by real-time PCR. On day 4 after infection, the virus load in the intestines of mice that received VHH1-anchored lactobacilli was significantly lower than that of mice given lactobacilli expressing an irrelevant anchored VHH (98.3% geometric mean reduction) (P<.05). The clearance rate of rotavirus (when no vp7 transcript was detectable) was 43% in the group that received VHH1-anchored lactobacilli (figure 6), compared with 17% in the untreated group

Because most naturally occurring rotavirus infections are caused by low doses of virus, we also tested the prophylactic potential of the VHH1-anchored lactobacilli for the prevention of infection and disease in mice challenged with a low inoculum of RRV (4 DD₅₀). The pups inoculated with RRV developed diarrhea, and, at day 2 after infection—a time point that marks the peak of clinical disease—the diarrhea prevalence was 50%, compared with 6% in the group treated with VHH1-anchored bacteria—an 88% reduction (P=.0015). Intestinal sections collected from the distal ileum of mice were also analyzed for virus load by real-time PCR. None of the mice treated with VHH1-anchored lactobacilli had any detectable virus on day 3, a reduction of 99.99% (geometric mean) relative to the untreated group (P<.05)

Discussion

In the present study, we have, to our knowledge for the first time, successfully expressed llama antibody fragments in lactobacilli. The VHHs that are formed by a single polypeptide exhibit several advantages over scFvs: they are smaller, are markedly more acid and heat resistant, and are much easier to express with an intact spatial structure. These properties make them suitable for the treatment of gastrointestinal infections

The generation and selection of VHH fragments was performed using a serotype G3 RRV strain [20]. The VHH1 fragment selected reacted with a variety of human strains but showed very limited cross-reactivity with the murine EDIM rotavirus strain, as tested by ELISA (unpublished data). We therefore chose the RRV mouse-infection model to test the activity of the lactobodies. Although it is a heterologous pathogen, the infection of mice with the simian SA11 rotavirus strain leads to pathophysiological alterations that are characteristic of rotavirus infection in humans [26 –28]

VHH1 is most likely directed against an outer coat protein, either VP4 or VP7, given that both immunization and selection were performed using the whole virus. SEM results also showed that the VHH1 expressed by these lactobacilli bound to intact virus particles. Attempts to map the epitope by direct methods have failed to date, which suggests that the antibody may recognize a conformational epitope

As was evident in the in vitro assays, VHH1 purified from the supernatant of VHH1-secreted lactobacilli was effective in neutralizing rotavirus. Nevertheless, only VHH1-anchored lactobacilli successfully reduced viral load, normalized pathological features, and mitigated diarrhea in our mouse model. In line with this, we recently observed that purified monovalent VHH1 fragments produced in yeast mitigated rotavirus-induced diarrhea in the mouse pup model but only when they were administered at very high doses (>10 μg daily) [20]. Although we used a strong constitutive promoter, VHH1-secreting lactobacilli cannot produce this amount of VHH1. When the anchored construct was used, however, the numerous antibody fragments expressed on the bacterial surface resulted in the formation of “biological beads” that allowed high-avidity binding due to multivalency and, thus, promoted strong agglutination and subsequent clearance of the virus. This is of particular importance in the gastrointestinal tract, where microbial attachment is often based on low-affinity carbohydrate-protein interactions for tight adhesion to mucosal surfaces and biofilms

If used in developing countries, the genetically modified lactobacilli would have to be freeze-dried and administered in oral rehydration solutions or soft drinks. The reconstituted freeze-dried VHH1-anchored lactobacilli were shown to be as protective against rotavirus-induced diarrhea as their fresh counterparts, which suggests that the antiviral activity is retained after freeze-drying

A limited number of controlled clinical trials have shown that certain strains of lactobacilli exhibit both prophylactic and therapeutic properties in acute viral gastroenteritis [29 –31]. This may be linked to their ability to survive the gut transit, adhere to mucosa, and transiently colonize the intestine, where a significantly enhanced local immune response is suggested by an increased production of secretory IgA and stabilization of the mucosal barrier [29 –31]. In our infection model, treatment with nontransformed L. paracasei induced a modest reduction in diarrhea prevalence, which supports this notion. Although the mechanism of action of probiotics is not fully understood, the prophylactic effect of antibody expressing Lactobacillus in our system probably stems from a synergistic effect between an antiviral activity of the bacteria and the action of the lactobodies. For example, the rotavirus bound on the surface of Lactobacillus transformants might theoretically be killed by the high local concentration of antiviral substances produced by the latter (e.g., lactic acid)

The advantages of using genetically modified lactobacilli include cost-efficient production, a long shelf life when they are lyophilized, simple logistics for distribution, and ease of administration. In situ production of VHH antibody fragments locally in the gastrointestinal tract of the host not only reduces the cost of purification but also circumvents the practical problem of the degradation of orally administered antibodies in the stomach. If increased avidity or multispecificity is the key to the efficient neutralization of RRV, it should theoretically be possible to improve our results by concomitant expression of anti-RRV VHH fragments of varying specificities on the surface of lactobacilli or by using a mixture of recombinant clones. The use of lactobacilli expressing VHH antibody fragments might be the prophylaxis of choice for children with malnutrition and immunodeficiency. Furthermore, Lactobacillus-expressing antibodies could reduce rotavirus transmission by capturing virus particles in the intestine and might thus be used for the control of nosocomial rotavirus outbreaks. New, biologically safe, contained expression systems in which the antibody gene, devoid of antibiotic selection markers, is integrated into the chromosome of lactic-acid bacteria are being developed in our laboratory at present [32]. Tailor-made lactobodies may thus represent a new and versatile system for the passive immunization at mucosal surfaces and may be of major medical importance, especially in developing countries