Results

Preparation of the phage surface display library

An independent expression library was constructed for each of the three Borrelia strains as described in materials and methods. The size of the libraries varied between 3.3x 10⁶ (B. afzelii) and 7.5x 10⁷ (B. burgdorferi s.s. and B. garinii). The diversity of the libraries was tested by PstI restriction analysis of 12 randomly picked clones of each library. All analyzed fragments showed different sizes, indicating a high diversity of the libraries.

Selective enrichment of phage binding human IgG

The three phage libraries were panned separately in single wells and in addition, a mixture of equal amounts of each Borrelia library was selected. The amount of phage added varied between 1.8x 10¹¹ and 1.6x 10^{1 2}. The number of phage eluted by pH shift after affinity cycle 5 was up to 9 fold higher than the number washed out in the last wash-fraction, indicating a selective enrichment of specific phage, despite there was no increase in the total number of eluted phage. The number of eluted phage in rounds two to five were between 5.4x 10⁴ and 2.1x 10⁶. There was no significant difference between different libraries during the course of affinity selection. Restriction analysis of selected clones after five cycles of biopanning showed specific enrichment as indicated by an increase in the number of phagemids carrying inserts of identical length in agarose gels. From 124 clones with insert, only 16 different insert sizes were identified.

Identification of proteins

From the 124 clones which contained an insert, two of each insert size were sequenced. Sequence analysis revealed nine distinct inserts showing a high degree of homology with proteins from the Borrelia genome. Six of the open reading frames encoded already described proteins, including BBK32, a novel borrelial protein which has recently been described as a useful antigen for serodiagnosis of LB. The other three inserts revealed predicted, not yet expressed, hypothetical proteins. Protein characteristics and GeneBank accession numbers are summarized in table 3.

In order to determine the number of clones which contained one of the 9 identified proteins, agarose gels of restriction enzyme analyses were evaluated. As shown in table 3 the insert of 71 clones could be matched with one of the 9 identified Borrelia proteins, since they possessed identical length in agarose gels. Of the remaining 53 clones, 50 contained inserts which could be attributed to Borrelia DNA-fragments

inserted in the wrong reading orientation, presumably coding for short peptides. These peptides may represent peptide mimitopes, which could be interesting candidates and will need further investigation. One clone could not be correlated with a published sequence, and two clones coded for a human open reading frame. This might represent cross-contaminations from other libraries (167) used in the same laboratory or sequences not present in the reference strain in the databases.

Production of recombinant antigens

The high level expression vector pQE30 was used for heterologous expression of the selected B. burgdorferi proteins in E. coli. From the identified proteins only flgL and ctc resulted in a sufficient expression rate (about 1 mg/l bacteria culture in LB-medium). In addition, OspC was expressed as a control protein for the ELISA. The purified proteins showed molecular weight consistent with their predicted molecular weight, calculated from the amino acid sequences. The purity and yield of these preparations was high enough to allow testing in ELISA to evaluate their potential for the serodiagnosis of LB.

Pilot evaluation of antigens for serodiagnosis

Antigen was coated to a microtiter plate and incubated with patient and control sera.

Furthermore, a Borrelia lysate mixture from all three human pathogenic strains (B.

garinii, B. afzelii, B. burgdorferi s.s.) was treated the same way to compare the data obtained. As shown in figure 1 patients sera showed significantly higher ELISA signals than sera from controls against the OspC protein, the lysate and the ctc protein. For the flgL protein, no difference between control and patient sera was detectable. However, compared to the other three proteins, the signals for the flgL protein were very low (patients 0.036 +/- 0.01; controls 0.019 +/- 0.01) and might reflect unspecific background. Since those signals were so low, background signals from the bacterial preparations in the other protein preparations can probably be neglected. The signal

intensity of the newly identified protein ctc (patients 0.248 +/- 0.05; controls 0.032 +/- 0.01) was comparable to the signal obtained for OspC (patients 0.229 +/- 0.03; controls 0.03 +/- 0.02) and the lysate (patients 0.326 +/- 0.04, controls 0.042 +/- 0.01). These data show that sera from LB patients contain antibodies against the ctc protein from Borrelia.

Table 3: List of the Borrelia genes and related products.

The open reading frames which were identified by sequencing the pJuFo vector after enrichment against the panning pool. The number of clones indicates how many of the clones analyzed carried the respective insert. The sequences were applied for an European patent (registration number 02 018 511.2)

OspC

patient control

0.00 0.25 0.50

***

A405

lysate

patient control

0.00 0.25 0.50

0.75

***

A405

ctc

patient control

0.00 0.25 0.50 1.00 1.25

**

A405

flgL

patient control

0.00 0.25 0.50

A405

Figure 1: Pilot evaluation of the proteins

ELISA signals of sera from 22 patients (Acrodermatitis chronica athropicans patients as circle, patients with neurological disorders as quad and patients with Arthritis as triangle) and 13 sera from healthy controls in a dilution of 1:200. Signals were expressed as A405 (absorbance at 405nm). Bound antibodies were detected using anti-human IgG labeled with alkaline phosphatase. Data from patient sera are expressed as individual results and group mean.

** and *** represent p values <0.01 and (0.001, respectively.