You don't say what database you are using, but I'll guess maybe it's MySql? Anyway, the following principles will apply to almost any relational database package (MS Access, SQL Server, PostGres etc)
For any table that has a relationship with the user table, you need to have a "foreign key", which is a field that relates back to another table. For example, taking your design:
User
int id;
varchar user_type;
varchar password;
varchar registrationDate;
Buyer:
int id;
int user_id;
varchar housesBought;
Seller
int id;
int user_id;
varchar rating;
varchar housesSold;
Note that I've changed the primary keys (id fields) to integers. You should not use varchar for numerical data, as its very inefficient when it comes to indexing.
The user_id fields link to the id field of the User table.
However, I would actually recommend a different design altogether. Your design, with separate tables for buyers and sellers attributes has some limitations. For example, what if someone is both a buyer and a seller - then you have data stored in two different places, which is unnecessary in this simple example. Also, the relationship that we have defined between the user table and the other two tables is a one-to-one relationship. For any given user, you'll always have only one record in each table. This implies that you might be better just storing all the information in one table.
Generally in relational database design, we use 1-to-many relationships as an efficient way to partition and link data. For example, let's say you have a table storing addresses for your users. Each user could have more than one address, so we have something like this:
Address
int id;
int user_id;
bool is_default_address;
varchar address_line1;
varchar address_line2;
varchar county;
varchar postcode;
Because we can have many addresses per one user, this is therefore a one-to-many relationship, using the foreign key user_id to link the tables.
Now, my final recommendation is to overhaul your database design to get away from those 1-to-1 relationships. I suggest you go for something like this:
User:
int userid;
varchar password;
varchar registrationDate;
int rating;
int housesSold;
int housesBought;
Property
int id;
int seller_id;
int buyer_id;
varchar description;
varchar address;
varchar postcode;
decimal price;
Now you have all user details stored in one table - yes there's a small amount of redundancy, but its way more flexible, and easier to retrieve the data you need, without your website having to make necessary table joins.
The fields seller_id and buyer_id in the Property table are both foreign keys linking to the User.id field. Your web app can then count the number of links to a particular user, and use that to autopopulate the housesSold and housesBought fields in the user table.
Relational database design is a huge topic, so I've only scratched the surface, but I hope these are enough pointers to get you going. It'd be worth investing in a good book, perhaps the relevant O'Reilly book for your chosen database platform.